FALL PREVENTION RESEARCH PAPER

RESEARCH

Nursing & Health Sciences Research Journal

Journal homepage: https://scholarlycommons.baptisthealth.net/nhsrj/

An Exploration of the Association of Patient Characteristics and

Pharmacological Treatments to Inpatient Falls among Patients At-risk for

Falling during Hospitalization

Julie David, Maria M. Ojeda, James O. Adefisoye, and Winifred Pardo

ABSTRACT

Introduction/Background: Falls may be the most commonly reported incidents in the acute care setting, and a

frequent cause of harm in the hospital. Studies have focused on identifying risk factors for falls and interventions

aimed at reducing the risk of falling. The purpose of this study was to describe and compare patient characteristics

and pharmacological treatments between patients who fell and patients who did not fall, among a sample of patients

deemed to be at-risk for falling during hospitalization. Additionally, the study aimed to identify independent predictors

of falls among patients at-risk for falls during hospitalization.

Methods: An observational, cross-sectional study involving the analysis of retrospective patient records. A

convenience sample of all patients with a Morse Fall Scale of >45 over a 1-year period, was extracted from electronic

medical records. Descriptive statistics of demographic characteristics and medication classes were generated to

compare those who fell to those who did not fall. To examine significant predictors of falls, logistic regression

(univariate and multivariable) were employed.

Results/Findings: The sample consisted of 4,978 valid patient records. White non-Hispanics constituted 60% of the

falls group but only 24% of the non-falls group. A larger proportion of those who fell received antiemetics or insulin

compared to those who did not fall. Univariate regression analysis found that race and 39 medication classes were

independently associated with falls. Multivariable regression analysis showed that race and 11 medication classes

were associated with the odds of falling.

Conclusions: White patients were more likely to fall than patients of other races. New associations were found

between the odds of falling and antiprotozoals, diagnostic agents, and gastrointestinal agents. Prospective studies are

needed to determine the predictive accuracy of these factors. Bedside practitioners should understand the mechanism

and onset of action of medications so that individualized safety precautions may be implemented. By including classes

of medications as part of fall-risk assessment, patient safety may be optimized and falls avoided in this high risk

population.

Keywords: falls, inpatient, hospital, patient characteristics, medications, pharmacological treatment

INTRODUCTION

Falls continue to present a major patient safety

concern both in the community and in hospitals. In

2015, direct medical costs for fatal falls were estimated

at $637.5 million and $31.3 billion for non-fatal injuries

(Burns, et al., 2016). Falls is one of the most frequent

causes of harm in the hospital. Bouldin et al., (2013)

reported a prevalence rate of 3.56 per 1000 patient days

and an injury rate of 0.93 per 1000 patient days. Thus,

regulatory agencies, such as the Centers for Medicare and Medicaid Services (CMS) and The Joint

7

David et al.: An Exploration of the Association of Patient Characteristics and

Published by Scholarly Commons @ Baptist Health South Florida, 2018

Commission (TJC), have placed great emphasis on fall

prevention programs and have included this as a

requirement for billing and accreditation (Quigley &

White, 2013).

Definition of Falls

The National Database of Nursing Quality

Indicators (NDNQI) in its 2016 guidelines, defined a

patient fall as an unplanned descent to the floor with or

without injury to the patient. NDNQI reports two fall

indicators (rate of total falls per 1000 patient-days and

the rate of falls with injury per 1000 patient-days),

which were endorsed by the National Quality Forum.

Factors Associated with Falls

Studies focused on the prevention of falls have

attempted to identify the factors that contribute to fall

risk (Deandrea, et al., 2013; Dykes, Hurley, Benoit, &

Middleton, 2009) as well as interventions (Dykes et al.,

2010) that may help reduce the risk for falling in the

acute care setting. Incorporating an externally validated

instrument, such as Morse Fall scale, into the electronic

medical record has been identified as an important

strategy in identifying patients at-risk for falling

(DuPree & Musheno, 2014). However, due to the

multifactorial nature of falls, even externally validated

tools have not been able to accurately assess the risk of

falling on a consistent basis (Williams, Szekendi &

Thomas, 2014).

Some studies have focused on identifying the

modifiable and non-modifiable risk factors that may

predict patient falls such as patient characteristics, fall

risk assessment, unassisted ambulation, medications,

bathroom-related/toileting, bed and chair alarms, call

light, environment and equipment, handoff communi-

cation, education, and change management (DuPree &

Musheno, 2014). Additionally, gait/balance deficit or

lower extremity problems, confusion, use of

sedatives/hypnotics, use of diabetes medications,

increasing patient-to-nurse ratio and activity level of

“up with assistance” compared with “bathroom

privileges” were found to be associated with increased

risk of falling (Krauss et al., 2005).

An exploratory research conducted by Tzeng and

Yin (2008) found that only 13.5% of falls examined

were medication-related falls. A more recent work

found that pain medications/opiates, cardiac

drugs/antihypertensives, sedatives/hypnotics, and anti-

psychotics/ antidepressants were commonly taken by

patients within the 24 hour period just prior to a fall

(Williams, Szekendi & Thomas, 2014). These findings

were similar to the study findings of Krauss and

colleagues (2005) in which nonnarcotic analgesics,

antiarrythmic agents, and sedatives/hypnotics were

associated with falls, in addition to diabetes

medications.

In summary, with the exception of the study by

Krauss et al. (2005) and Anderson, Dolansky, Damato,

and Jones (2014), most studies have been descriptive in

nature and no attempt was made to determine what

factors actually predicted the occurrence of falls among

hospitalized patients (DuPree & Musheno, 2014; Tzeng

& Yin, 2008; Williams, Szekendi, & Thomas, 2014).

In Krauss et al. (2005), patients who fell were compared

to patients who did not fall during hospitalization on a

variety of potential predictors. The patients included in

the study were assessed to be at varying levels of risk

for falling on admission, therefore it is difficult to

determine to what extent pre-existing risk may have

confounded or contributed to the findings of the study.

A more recent study conducted by Anderson et al.

(2014) indicated that male gender and age greater than

65 were predictive of serious fall injury among

hospitalized patients. Yet, there has been little research

focusing on subgroups of patients at varying levels of

risk for falling or the identification of factors associated

with the occurrence of falls within such subgroups.

The present study focused on patients assessed to

be at high risk for falling during hospitalization. The

characteristics of those who actually fell were

compared to those who did not fall and significant

predictors were identified.

The study aims were to:

1. Describe the distribution of certain patient characteristics and pharmacological treatments

among patients at-risk for falling during

hospitalization.

2. Compare those who actually fell to those who were at-risk but did not fall during

hospitalization for significant differences in

patient characteristics and prescribed

pharmacological treatments.

3. Determine the association of patient characteristics to the occurrence of falls

among patients at-risk for falling during

hospitalization.

4. Determine the association of pharmacological treatments to the occurrence of falls among

patients at-risk for falling during

hospitalization.

5. Identify independent predictors of falls among patients at-risk for falls during

hospitalization.

METHODS

Institutional review board approval for the study

was obtained in September of 2015. A waiver of

informed consent was obtained since the study was of

minimal risk and did not involve identifiable patient

information, interventions or changes in patient care.

8

Nursing & Health Sciences Research Journal, Vol. 1, Iss. 1 [2018], Pg. 7-18

https://scholarlycommons.baptisthealth.net/nhsrj/vol1/iss1/4

Design

A cross-sectional observational design was utilized

to examine similarities and differences between

patients who fell and patients who did not fall during

their period of hospitalization. All medical records data

included in the study were collected retrospectively.

Setting

The study was conducted at a 148-bed community

hospital in southeastern United States. The patient

population was racially and ethnically diverse;

according to the U.S. Census Bureau (2010), the

surrounding community is approximately 58%

Hispanic, 31% Black, 10% White, and 1% other races.

The hospital’s unique location, surrounded by farmland

but only 40 miles from a major metropolitan city

allowed it to serve a mix of both rural and urban

populations.

Sample

A convenience sample of 5,356 records from

inpatients deemed to be at high-risk for falling at any

point during hospitalization between July 1st, 2014 and

June 30th, 2015 were included in the study. Patient-

level data was extracted from electronic medical

records, as well as administrative and clinical pharmacy

databases. A pseudo identification number was

assigned to each patient record. Medical record

numbers were not retained in the study database.

Patient records were included in the study if they had a

documented Morse Fall Scale score of 45 or greater,

indicating high-risk for falling at any point during

hospitalization. Excluded from the study were

outpatients and the records of patients with all recorded

Morse Fall Scale scores of less than 45, no documented

Morse Fall Scale score, and those with no Morse Fall

Scale score documented prior to falling. After

screening for inclusion and exclusion criteria, and the

removal of 378 patient records with unrealistic or

missing height and weight measurements, a total of

4,978 records were found eligible for the study.

Measures

Variables representing patient characteristics were

selected based on the extant literature linking certain

patient characteristics to the occurrence of falls during

hospitalization and the researchers’ determination of

factor’s within the hospital’s unique patient population

that may have contributed to falls. Patient

characteristics measured on a continuous scale were

age, height, and weight; categorical measures included

patient gender, race, preferred language, insurance

status, history of smoking, history of alcohol use, and

primary admission diagnosis. In addition, the impact

of patient comorbidities on the occurrence of falls

during hospitalization was assessed using the Charlson

Comorbidity Index (CCI). The CCI is a weighted index

of 19 patient comorbidities, with demonstrated

predictive validity for several hospital outcomes

including complications, length of stay, mortality

(Charlson, Szatrowski, Peterson & Gold, 1994;

Johnson et al., 2015; Schmolders et al., 2015). By

determining the presence or absence of the included

comorbidities and adding their associated weights

(which range between 1 and 6 points), a risk score is

generated; higher scores indicate higher risk for

mortality. Since its inception, the CCI has undergone

several revisions, in this study scoring of the CCI was

based on an updated version of weights and

International Classification of Diseases 9th Edition

(ICD-9) diagnostic codes representing patient

comorbidities (Quan et al., 2005; Quan et al., 2011).

Variables representing pharmacological treatments

were coded based upon major drug classification.

Since the initial data was received in the form of

generic drug names or brand names, members of the

research team recoded each drug into their major drug

classifications in accordance with the American

Hospital Formulary Service (AHFS) – 2016 Drug

Formulary and under the guidance and discretion of a

research team member with pharmacological expertise.

A total of 131 major drug classifications were

represented within the sample.

Analysis

Analyses were conducted using RStudio 0.97.551

(R Studio, 2013) incorporating R 3.2.1 (R Core Team,

2015) open source statistical analysis software.

Descriptive statistics using frequencies, percentages,

and measures of central tendency were generated for

the overall sample and subgroups of patients who fell

and did not fall (study aims “1” & “2”). Univariate

logistic regression analysis was conducted to identify

factors associated with patient falls (study aims “3” &

“4”); variables with a p < .20 were considered

significant and were retained for multivariable

modeling. Multivariable logistic regression using

backward elimination was conducted to identify

independent predictors of patient falls (study aim “5”);

variables with a p < .10 were retained and used to build

the final model.

RESULTS

Of the 4978 records included in the analysis, only

40 (0.80%) patients actually fell during their period of

hospitalization while the remaining 4,938 did not fall.

Table 1 provides a complete description of the sample

demographics. Eight of the ten most frequently

occurring medication classifications were similar for

the overall sample, the subgroup that did not fall, and

the subgroup that fell (Table 2). The administration of

9

David et al.: An Exploration of the Association of Patient Characteristics and

Published by Scholarly Commons @ Baptist Health South Florida, 2018

those medications range from 40.8% to 83.2% of the

sample. However, antiemetics and insulins were found

to be among the ten most frequent medication classes

for those who fell but not among those who did not fall.

Table 1

Subgroup Comparison based on Fall Status (Did Not Fall, Fell) and Description of the Overall Sample (N=4978)

Did not Fall (n = 4938) Fell (n = 40) Overall sample (n = 4978)

n (%) M (sd) n (%) M (sd) n (%) M (sd)

Age -- 61.8 (18.9) -- 59.5 (16.4) -- 61.8 (18.9)

Gender

Male

Female

2272 (46%)

2666 (54%)

19 (47.5%)

21 (52.5%)

2291 (46%)

2687 (54%)

Race

White Hispanic

White Non-Hispanic

Black/African American

Black Hispanic

Other

Asian/Pacific Islander

2279 (46.15%)

1195 (24.20%)

1112 (22.52%)

188 (3.81%)

131 (2.65%)

33 (0.67%)

7 (17.5%)

24 (60%)

8 (20%)

1 (2.5%)

0 (0%)

0 (0%)

2286 (45.92%)

1219 (24.49%)

1120 (22.50%)

189 (3.80%)

131 (2.63%)

33 (0.66%)

Preferred Language

English

Spanish

Creole

Other

3918 (79.34%)

966 (19.56%)

33 (0.67%)

21 (0.43%)

37 (92.5%)

2 (5.00%)

1 (2.50%)

0 (0.00%)

3955 (79.45%)

968 (19.45%)

34 (0.68%)

21 (1.42%)

Insurance Status

Insured

Self Pay

4651 (94%)

287 (6%)

38 (95%)

2 (5%)

4689 (94%)

289 (6%)

Smoker

Yes

No

1768 (35.8%)

3170 (64.2%)

19 (47.5%)

21 (52.5%)

1787 (35.9%)

3191 (64.1%)

Alcohol Use

Yes

No

1094 (22.2%)

3844 (77.8%)

9 (22.5%)

31 (77.5%)

1103(22.2%)

3875 (77.8%)

CCIa

0

1

2

3

4

5

6

7

8

9

10

11

12

13

2405 (48.7%)

770 (15.6%)

635 (12.8%)

348 (7%)

319 (6.6%)

129 (2.6%)

164 (3.3%)

77 (1.6%)

47 (0.9%)

23 (0.5%)

15 (0.3%)

3 (0.05%)

1 (0.02%)

2 (0.03%)

18 (45%)

9 (22.5%)

3 (7.5%)

3 (7.5%)

5 (12.5%)

0 (0%)

1 (2.5%)

0 (0%)

1 (2.5%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

2423 (48.7%)

779 (15.6%)

638 (12.8%)

351 (7.1%)

324 (6.5%)

129 (2.6%)

165 (3.3%)

77 (1.5%)

48 (1%)

23 (0.5%)

15 (0.3%)

3 (0.05%)

1 (0.02%)

2 (0.03%)

Note. aCCI = Charlson Comorbidity Index score.

10

Nursing & Health Sciences Research Journal, Vol. 1, Iss. 1 [2018], Pg. 7-18

https://scholarlycommons.baptisthealth.net/nhsrj/vol1/iss1/4

Table 2

Description of Top Ten Medication Classifications by Subgroup and for the Overall Sample (N = 4978)

Did not Fall (n = 4938) Fell (n = 40) Overall Sample (n = 4978)

Medication Classification n (%) Medication Classification n (%) Medication Classification n (%)

Replacement Preparations 4143 (83.9%) Replacement Preparations 37 (92.5%) Replacement Preparations 4143 (83.2%)

Anticoagulants 3724 (75.4%) Anticoagulants 31 (77.5%) Anticoagulants 3724 (74.8%)

Proton-pump Inhibitors 3350 (67.8%) Proton-pump Inhibitors 30 (75.0%) Proton-pump Inhibitors 3350 (67.3%)

Antibacterials 3144 (63.7%) Antibacterials 30 (75.0%) Antibacterials 3144 (63.2%)

Opiate Agonists 2889 (58.5%) Opiate Agonists 29 (72.5%) Opiate Agonists 2889 (58.0%)

NSAIDSa 2145 (43.4%) Anticonvulsants 26 (65.0%) NSAIDSa 2145 (43.1%)

Benzodiazepines 2141 (43.4%) Benzodiazepines 26 (65.0%) Benzodiazapines 2141 (43.0%)

Beta-Adrenergic Blockers 2103 (42.6%) Antiemetics 24 (60.0%) Beta-Adrenergic Blockers 2103 (42.3%)

Anticonvulsants 2082 (42.2%) Analgesics/Antipyretics 21 (52.5%) Anticonvulsants 2082 (41.8%)

Analgesics/Antipyretics 2033 (41.2%) Insulins 19 (47.5%) Analgesics/Antipyretics 2033 (40.8%)

Note. aNSAIDS = Non-steroidal anti-inflammatory drugs.

Results of Univariate Regression

A total of 142 predictor variables were available to be

regressed against fall status, including 131 variables

representing a variety of major medication classifications.

Univariate logistic regression found 40 significant (p <

.20) predictors of patient falls. From these results, only

race was retained among the demographic variables,

the remaining 39 were representative of major

medication classifications. These results are presented

in Table 3.

Results of Multivariable Regression Analysis

Prior to multivariable regression, racial categories

were collapsed into three groups (“White Hispanic”,

“Black/African American”, “White-Non Hispanic and

Other”). We created the “White Non-Hispanic and

Others” category from the “White”, “Asian/Pacific

Islander”, “Black Hispanic” and “Other” race

categories since there was no significant difference

among these categories at the 5% significance level

based on results from the univariate regression. The

“White Non-Hispanic and Others” category was then

used as reference category for race in the multivariable

model. The multivariable model was then constructed

which included race and the 39 medication classes from

the univariate model. The results of the multivariable

analysis are presented in Table 4.

Considering the limited number of fall events and

a high number of redundant medication classes found

in the results (Table 4), we desired to reduce this model

further. We built a new multivariable logistic

regression model and excluded medication classes that

did not have a p-value < .10 from the previous model.

The final model is presented in Table 5. The Likelihood

Ratio Test (LRT) indicated that the overall model was

significant [χ2(13) = 78.31, p = .000]. Each of the 12

variables included in the model were significant

predictors of fall status. The odds of falling for the

Black/African American group was almost 3 times

lower, and the White Hispanic group was more than 6 times lower compared to patients of White-Non

Hispanic and Other races. Most of the medication classes

than 13 times lower odds of falling. Patients who were

administered gastrointestinal drugs-miscellaneous

classification had odds of falling that were more than

49 times higher, barbiturates more than 10 times higher,

diagnostic agents 6 times higher, centrally acting

skeletal muscle relaxants by almost 5 times higher than

those who were not. In addition, patients who were

administered antiparkinsonian agents had an odds of

falling about 4.5 times higher, general anesthetics by

about 4 times higher, thiazide diuretics by more than 3

times higher, local anesthetics by almost 3 times higher,

autonomic drugs-miscellaneous by about 2.6 times

higher, and antiemetics 2.4 times higher than those who

were not.

DISCUSSION

In this study, the investigators observed that White

Hispanics and Black/African Americans had lower

odds of falling compared to other races. While White

Hispanics and Black/African Americans together

constituted about 68% of the sample, White Non-

Hispanics were only 24% of the sample but constituted

60% of the group that experienced falls. Most of the

previous studies looked at the association of other

patient characteristics such as age and gender

(Deandrea et al., 2016; Williams, Szekendi, & Thomas,

2014) to the occurrence of falls. There was a dearth of

information in the literature regarding the relationship

of race to the occurrence of falls except for an earlier

study conducted in the community that identified White

race as a risk factor for a serious fall-related injury

(Tinetti, Doucette, Claus, & Marottoli, 1995). This

earlier finding was similar to what we found in our

study of which the group containing White Non-

Hispanics had higher odds of falling. We were not

11

David et al.: An Exploration of the Association of Patient Characteristics and

Published by Scholarly Commons @ Baptist Health South Florida, 2018

Table 3

Results of Univariate Analysis - Logistic Regression results of Fall Status vs. each Predictor (N = 4978)

Predictor β SE(β) Z p

Racea

Asian/Pacific Islander -15.658 1872.033 -0.008 .993

Black Hispanic -1.329 1.024 -1.298 .194

Black/African American -1.027 0.410 -2.502 .012

Other -15.653 939.583 -0.017 .987

White Hispanic -1.878 0.431 -4.356 .000

Medication Classb

Alpha and Beta Adrenergic Agonists 0.952 0.532 1.790 .073

Ammonia Detoxicants 0.888 0.381 2.330 .020

Analgesics and Antipyretics 0.475 0.318 1.490 .140

Angiotensin II Receptor Antagonists 0.545 0.398 1.370 .170

Antibacterials 0.564 0.366 1.540 .120

Anticonvulsants 0.957 0.333 2.880 .004

Antidepressants 0.833 0.333 2.500 .013

Antiemetics 0.879 0.324 2.710 .007

Antimuscarinics Antispasmodics 0.727 0.446 1.630 .100

Antiparkinsonian Agents 1.509 0.610 2.470 .013

Antiprotozoals -1.766 1.014 -1.740 .081

Antitussives 0.747 0.482 1.550 .120

Antivirals 0.907 0.532 1.710 .088

Anxiolytics Sedatives and Hypnotics 0.877 0.398 2.200 .028

Autonomic Drugs Miscellaneous 1.092 0.420 2.600 .009

Barbiturates 1.658 1.033 1.600 .110

Benzodiazepines 0.904 0.333 2.720 .007

Blood Derivatives 0.693 0.531 1.300 .190

Caloric Agents 0.662 0.446 1.480 .140

Carbonic Anhydrase Inhibitors 1.836 0.614 2.990 .003

Cathartics and Laxatives 0.511 0.321 1.590 .110

Central Alpha Agonists 0.585 0.446 1.310 .190

Centrally Acting Skeletal Muscle Relaxants 1.539 0.535 2.870 .004

Diagnostic Agents 1.642 0.536 3.060 .002

General Anesthetics 1.380 0.448 3.080 .002

GI Drugs Miscellaneous 4.148 1.235 3.360 .001

Glycogenolytic Agents 1.503 1.030 1.460 .140

Immunosuppressive Agents 1.658 1.033 1.600 .110

Insulins 0.457 0.318 1.440 .150

Local Anesthetics 1.220 0.420 2.900 .004

Neuromuscular Blocking Agents 1.540 1.031 1.490 .140

Opiate Agonists 0.650 0.355 1.830 .067

Prokinetic Agents 0.847 0.482 1.757 .079

Replacement Preparations 0.916 0.602 1.523 .128

Serums 3.048 1.092 2.791 .005

Somatostatin Agonists 0.931 0.606 1.536 .125

Thiazide Diuretics 1.060 0.532 1.991 .046

Vaccines 0.548 0.367 1.492 .136

Vitamins 0.938 0.325 2.889 .004

Note. Criteria for statistical significance: p < .20: aComparison group is White Non-Hispanic. bComparison group is absent

(omitted from the table). The coding for the medication classes was dichotomous: 1 = present, 0 = absent.

12

Nursing & Health Sciences Research Journal, Vol. 1, Iss. 1 [2018], Pg. 7-18

https://scholarlycommons.baptisthealth.net/nhsrj/vol1/iss1/4

Table 4

Multivariable Logistic Regression Results – Predictors of Fall Status

Predictor β SE(β) Z p

Constant -5.848 0.694 -8.429 .000

Racea

Black/African American -0.818 0.435 -1.883 .060*

White Hispanic -1.849 0.474 -3.897 .000*

Medication Classb

Alpha and Beta Adrenergic Agonists -0.546 0.788 -0.692 .489

Ammonia Detoxicants -0.035 0.477 -0.073 .942

Analgesics and Antipyretics -0.043 0.370 -0.117 .907

Angiotensin II Receptor Antagonists 0.360 0.448 0.802 .422

Antibacterials 0.082 0.420 0.195 .845

Anticonvulsants 0.337 0.380 0.887 .375

Antidepressants 0.483 0.369 1.308 .191

Antiemetics 0.743 0.384 1.934 .053*

Antimuscarinics Antispasmodics -0.272 0.541 -0.502 .616

Antiparkinsonian Agents 1.532 0.691 2.217 .027*

Antiprotozoals -2.882 1.106 -2.606 .009*

Antitussives 0.454 0.544 0.834 .404

Antivirals 0.484 0.606 0.799 .424

Anxiolytics Sedatives and Hypnotics 0.529 0.445 1.187 .235

Autonomic Drugs Miscellaneous 0.867 0.467 1.857 .063*

Barbiturates 2.152 1.150 1.872 .061*

Benzodiazepines 0.157 0.382 0.411 .681

Blood Derivatives 0.004 0.659 0.006 .995

Caloric Agents 0.039 0.593 0.065 .948

Carbonic Anhydrase Inhibitors 1.080 0.758 1.424 .154

Cathartics and Laxatives -0.305 0.389 -0.784 .433

Central Alpha Agonists 0.115 0.484 0.238 .812

Centrally Acting Skeletal Muscle Relaxants 1.370 0.595 2.300 .021*

Diagnostic Agents 2.021 0.674 2.999 .003*

General Anesthetics 1.295 0.669 1.936 .053*

GI Drugs Miscellaneous 4.034 1.563 2.582 .010*

Glycogenolytic Agents 0.704 1.220 0.577 .564

Immunosuppressive Agents 1.360 1.176 1.156 .248

Insulins 0.101 0.390 0.259 .795

Local Anesthetics 0.841 0.477 1.764 .078*

Neuromuscular Blocking Agents 0.157 1.314 0.120 .905

Opiate Agonists -0.083 0.419 -0.197 .844

Prokinetic Agents 0.411 0.552 0.745 .456

Replacement Preparations 0.178 0.636 0.280 .780

Serums 2.140 1.316 1.626 .104

Somatostatin Agonists 0.188 0.712 0.263 .792

Thiazide Diuretics 1.122 0.608 1.847 .065*

Vaccines 0.207 0.404 0.513 .608

Vitamins 0.374 0.377 0.991 .322

Note. *Significant at p < .10. aComparison group is “White Non-Hispanic and Others”. bComparison group is absent

(omitted from the table). The coding for the medication classes was dichotomous: 1 = present, 0 = absent

13

David et al.: An Exploration of the Association of Patient Characteristics and

Published by Scholarly Commons @ Baptist Health South Florida, 2018

Table 5

Multivariable Logistic Regression Results

Predictor β SE(β) Z p OR [95% CI] Constant -5.052 1.376 -15.840 0.000 0.006 [0.003 – 0.011]

Racea

Black/African American -0.816 1.518 -1.953 0.051* 0.442 [0.183 – 0.962]

White Hispanic -1.820 1.580 -3.977 0.000 0.162 [0.061 – 0.374]

Medication Classificationb

Antiemetics 0.856 1.408 2.499 0.012 2.353 [1.209 – 4.675]

Antiparkinsonian Agents 1.512 1.916 2.325 0.020 4.536 [1.017 – 14.115]

Antiprotozoals -2.604 2.844 -2.492 0.013 0.074 [0.004 – 0.372]

Autonomic Drugs Miscellaneous 0.949 1.553 2.154 0.031 2.583 [1.006 – 0.795]

Barbiturates 2.349 3.006 2.134 0.033 10.474 [0.54 – 62.241]

Centrally Acting Skeletal Muscle Relaxants 1.531 1.745 2.750 0.006 4.621 [1.323 – 12.398]

Diagnostic Agents 1.806 1.869 2.887 0.004 6.086 [1.521 – 18.744]

General Anesthetics 1.432 1.601 3.043 0.002 4.187 [1.513 – 9.873]

GI Drugs Miscellaneous 3.896 4.814 2.479 0.013 49.185 [1.401 – 983.747]

Local Anesthetics 1.013 1.550 2.311 0.021 2.753 [1.077 – 6.15]

Thiazide Diuretics 1.205 1.800 2.050 0.040 3.337 [0.878 – 9.357]

Note. Criteria for statistical significance: p < .05. *Significant at p < .05 as one of the levels of the Race variable. aComparison

group is “White Non-Hispanic and Others”. bComparison group is absent (omitted from the table). The coding for the

medication classes was dichotomous: 1 = present, 0 = absent.

surprised by some of the results regarding the impact of

pharmaceutical agents. A substantial number of drugs

have been previously identified in the Beers (American

Geriatrics Society, 2015) list to have an association

with the incidence of falls. We noted that a larger

proportion of those who fell received antiemetics or

insulin compared to those who did not fall. These

medications are listed in the Beers criteria for

potentially inappropriate medication use in older adults

(American Geriatrics Society, 2015). In our study,

patients who took antiemetics had 2.4 times higher

odds of falling.

Thiazide diuretics were observed to be associated

with increased odds of falling. Thiazides have been

associated with falls particularly among women older

than 70 years of age. It is suspected that a common side

effect of these drugs, hyponatremia, is responsible

(Hwang & Kim, 2010). Patients with symptomatic

thiazide-induced hyponatremia tend to experience

muscle weakness, fatigue, and loss of energy which

may contribute to falls (Liamis, Filippatos, & Elisaf,

2016; Sardar & Eilbert, 2015). While hyponatremia

tends to occur most frequently on initiation of therapy,

it may occur at any time during treatment as a

consequence of physiological or environmental

changes (Liamis et al., 2016).

The study results also indicated that barbiturates

were associated with increased odds of falling.

Barbiturates have strong sedative-hypnotic effects even

at low doses, therefore, causing the following side

effects: drowsiness, lethargy, dizziness (American

Geriatrics Society, 2012). This finding was consistent

with that of Dauphinot et al. (2014) who indicated that

increased exposure to anticholinergic and sedative

medications during hospital stay is associated with

increased risk of inpatient falls. Similarly, Krauss et al.

(2005) found that the use of sedative-hypnotics was a

predictor that significantly increased a patient’s risk of

falling in the hospital. The study finding that

administration of central skeletal muscle relaxants

(SMR) was associated with falls was not surprising.

These drugs are anticholinergic agents and have

documented side effects such as confusion and

sedation, which may lead to falls.

Antiparkinsonian drugs also increased the odds of

falling. Many of the antiparkinsonian agents such as

Carbidopa-Levodopa may cause dizziness and

confusion which may contribute to the occurrence of

falls. However, the symptoms of Parkinson's disease

itself including muscular stiffness, freezing, shuffling

gait, balance impairment or stooped posture make it

difficult to discern to what extent the antiparkinsonian

drugs may have contributed to the observed incidence

of falls (Chen & Swope, 2014). It has been noted in the

literature, that gait/balance impairment or lower

extremity problem was another predictor that

significantly increased a patient’s risk for falling

(Krauss et al., 2005).

Our study revealed that local and general

anesthetic agents were associated with increased odds

of falling for patients. Interestingly, the study of

Anderson et al. (2014) showed that surgical patients

were statistically more likely than medical patients to

sustain a serious fall injury when the category “no

14

Nursing & Health Sciences Research Journal, Vol. 1, Iss. 1 [2018], Pg. 7-18

https://scholarlycommons.baptisthealth.net/nhsrj/vol1/iss1/4

injury” was removed from the analysis. Although the

research regarding postoperative falls is limited and

overall incidence is low at 1.6 cases per 10,000 patients

(Lam et al., 2016), we may expect falls in this

population because the residual pharmacologic and

neuromuscular blocking effects of anesthetic agents

may predispose patients to falls. Patients who receive

anesthesia, especially surgical patients, receive a

combination of agents that have synergistic effects on

the central nervous system (CNS), neuromuscular and

GI system that can last for several hours or even days.

It is worth mentioning that Lam et al. (2016) observed

that post-anesthesia falls occurred at the bedside

usually during the day with the presence of clinicians

and after the first uneventful getting out of bed.

Although local anesthetic agents such as lidocaine

patch are not fully systemically absorbed, CNS side

effects such as fainting, dizziness, weakness,

confusion, and blurred vision are still reported

(Drasner, 2015).

Our study revealed a previously unreported link

between administration of diagnostic agents and an

increased odds of falling. We did not find any prior

studies associating these drug category to patient falls.

However, some of the diagnostic agents such as

diatrizoate meglumine and diatrizoate sodium, have

high osmolality and are water-soluble (Salix

Pharmaceuticals, 2016). This may lead to electrolyte

imbalances, diarrhea and dehydration resulting in

hypovolemia and hypotension. This is interesting to

note because based on this hospital’s previous analysis

of post-fall data, the most common reason why patients

fell was toileting. This was consistent with a large scale

study conducted by Williams et al. (2014) who noted

that toileting constituted 23% of common patient

activities at the time of a fall. Special care should be

taken to adequately hydrate patients before the

administration of some diagnostic agents to prevent

dehydration and to offer toileting frequently after the

procedure.

Another new finding was the association of

methylnaltrexone with the incidence of falls.

Methynaltrexone is a Mu opioid antagonist that is

indicated for the treatment of opioid-induced

constipation (Salix Pharmaceuticals, 2016). It works

by reducing the constipating effects of opioids without

impacting the analgesic effects of opioids (Salix

Pharmaceuticals, 2016). Because methylnaltrexone

does not cross the blood-brain barrier, we did not

expect it to have any impact on falls (Salix

Pharmaceuticals, 2016). However, we have surmised

that this is more likely to be the result of its rapid onset

of action rather than its mechanism of action. Studies

have suggested that 30% of patients on

methylnaltrexone have a bowel movement within 30

minutes of the first dose (Slatkin et al., 2009; Thomas

et al., 2008). One of its known side effects is diarrhea.

It is possible that falls may occur among some patients

who suddenly feel the need for toileting and in the

emergent rush to the bathroom for relief fail to call for

assistance from staff.

An interesting finding in this study was that

antiprotozoal agents were associated with decreased

odds for falling for the patients. The vast majority of

the medications administered in this category was

metronidazole given intravenously or orally. This

medication is frequently administered to patients with

protozoal and bacterial infections. It is also usually

given for certain gastrointestinal infections such as

amoebeasis, intra-abdominal infection, Clostridium

difficile-related diarrhea, Chron’s disease and

Helicobacter pylori (American Society of Hospital

Pharmacists, 2016). This was the only drug category

that lowered the patient’s odds of falling. Perhaps, since

these medications were given to patients with

gastrointestinal symptoms, certain precautions have

already been implemented and more vigilance was

offered by the staff to anticipate and meet toileting

needs.

Strengths

This is the first study that we know of to examine

fall-related outcomes among patients deemed at high-

risk for falling in the hospital setting. It was performed

using a large sample of 4,978 unique medical records.

Unlike previous studies, we did not include

environmental factors (e.g. raised bedrails, non-slip

footwear, and floor type) because of the ubiquity in the

implementation of such safety measures within

hospitals. By focusing on non-modifiable patient

characteristics and pharmaceutical treatments, we

identified a parsimonious list of 12 factors associated

with falls among high-risk patients. Thus, we expect

these findings will be amenable for follow-up

validation studies and eventual translation to the

practice setting.

Limitations

This study was conducted at a single hospital,

therefore the findings may not be readily generalized to

other populations. As with all observational research,

our results should not be interpreted to indicate

causation but rather, association. Further, the accuracy

of the results of studies based on medical records data

are contingent upon the accuracy and completeness of

the medical record and its abstraction. In addition,

during the year covered by our study the facility had a

low incidence of falls and a large number of patients

deemed at high-risk for falling, this resulted in an

unbalanced sample. We excluded the records of

patients who did not receive a Morse Fall Scale score

prior to falling, we cannot say for certain that such

15

David et al.: An Exploration of the Association of Patient Characteristics and

Published by Scholarly Commons @ Baptist Health South Florida, 2018

patients were not different in some way from those who

were scored.

Implications

Certain medications are known risk factors for

falls, which the literature and the results of this study

support. Nevertheless, there were certain medication

classes identified in this research as predictors for falls

that were not previously identified in the literature. It

is important to assure that the bedside clinician knows

the mechanism and onset of action of these medications

and implement safety precautions towards falls

prevention. For example, certain gastrointestinal

medications have a rapid onset of action which

necessitates that toilet facilities should be nearby

immediately following administration. Additionally,

assistance with toileting should also be immediately

available to patients so as to minimize risk for falls on

their way to the bathroom. It would further enhance

patient safety, if patients taking general and even local

anesthetic medications are flagged so that fall risk

reassessment can be conducted more frequently. This

is particularly important in light of recommendations

that post-surgical patients are to be encouraged to get

out-of-bed as early as possible in avoidance of post-

operative pneumonia and other complications. The

surveillance for fall risk and assistance with ambulation

should still be offered after the first successful

ambulation when patients could still be experiencing

the residual effects of the anesthetic agents.

The results of this study indicate the need for future

research regarding predictors for falls among high-risk

patients. Replication studies with more balanced

sample sizes may shed light on the consistency of our

findings. Further, there is an opportunity to identify

subpopulations of patients who are at highest risk by

examining how necessary treatments may interact and

contribute to the occurrence of falls during

hospitalization.

CONCLUSION

Some hospitalized patients fall despite the best

efforts of individual clinicians and organizations to

mitigate risk within the care environment. Thus,

preventing falls will continue to be a patient safety

imperative in the acute care setting. Due to the

multifactorial causation of falls, hospitals may need to

implement several approaches in order to adequately

address the problem. Knowledge regarding non-

modifiable patient characteristics and necessary

pharmacological treatments that predispose patients for

falling are an invaluable addition to our understanding

of factors contributing to patient fall-risk as well as the

improved alignment of safety precautions based on

individual patient’s needs.

DECLARATION OF INTEREST

The authors whose names are listed below have

indicated that they have no affiliations with or

involvement in any organization or entity with any

financial interest in the subject matter or materials

discussed in this manuscript.

AUTHORS

Julie David, MSN, ARNP, ANP-BC, Director,

Magnet Project and Advanced Practice,

Homestead Hospital, Homestead, FL, US. Correspond-

ence regarding this paper can be directed at

julieda@baptisthealth.net.

Maria M. Ojeda, DNP/PhD, MPH, BA, ARNP,

NP-C, BC-ADM, Nurse Scientist, Homestead Hospital,

Baptist Health South Florida, Miami, FL, US.

James O. Adefisoye, MS, Statistician, Homestead

Hospital, Homestead, FL, US.

Winifred Pardo, BS, PharmD, BCPS, Clinical

Pharmacy Supervisor and Residency Program Director,

Homestead Hospital, Homestead, FL, US.

REFERENCES

American Geriatrics Society. (2015). Beers Criteria Update

Expert Panel - American Geriatrics Society: 2015

updated Beers criteria for potentially inappropriate

medication use in older adults. Journal of the

American Geriatrics Society, 63(11), 2227-2246.

doi: 10.1111/jgs.13702. PubMed PMID: 26446832.

American Geriatrics Society. (2012). Beers Criteria Update

Expert Panel - American Geriatrics Society: 2012

updated Beers criteria for potentially inappropriate

medication use in older adults. Journal of the

American Geriatrics Society, 60(4), 616-631 doi:

10.1111/j.1532-5415.

American Society of Hospital Pharmacists. (2016). AHFS

drug information – 2016 Edition. Published by

authority of the Board of Directors of the American

Society of Hospital Pharmacists.

Anderson, C., Dolansky, M., Damato, E., & Jones, K. (2014).

Predictors of serious injury in hospitalized patients.

Clinical Nursing Research 24(3), 269-283.

Bouldin, E., Andresen, E., Dunton, N., Simon, M., Waters,

T., Liu, M., Daniels, J. Mion, L., & Shorr, R.

(2013). Falls among Adult Patients Hospitalized in

the United States: Prevalence and Trends. Journal

of Patient Safety, 9(1), 13-17.

Burns E., Stevens J., & Lee, R. 2016. The direct costs of fatal

and non-fatal falls among older adults—United

States. Journal of Safety Research. Sep(58), 99-

103.

Charlson, M., Szatrowski, T. P., Peterson, J., & Gold, J.

(1994). Validation of a combined comorbidity

16

Nursing & Health Sciences Research Journal, Vol. 1, Iss. 1 [2018], Pg. 7-18

https://scholarlycommons.baptisthealth.net/nhsrj/vol1/iss1/4

index. Journal of Clinical Epidemiology, 47(11),

1245 – 1251.

Chen J.J., & Swope D.M. (2014). In: Dipiro J.T. et al (eds),

Pharmacotherapy: A Pathophysiology Approach

(9 ed). New York, NY: McGraw-Hill.

Dauphinot, V., Fuare, R., Omrani, S., Goutelle, S., Bour-

guignon, L., Krolak-Salmon, P., & Mouchoux, C.

(2014). Exposure to anticholinergic and sedative

drugs, risk of falls, and mortality – an elderly

inpatient, multicenter Cohort. Journal of Clinical

Psychopharmacology, 34(5), 565-570.

Deandra, S., Bravi, F., Turati, F., Lucenforte, E., La Vecchia,

C., & Negri, E. (2012). Risk Factors for fall in older

people in nursing homes and hospitals. A

systematic review and meta-analysis. Archives of

Gerontology and Geriatrics. 56(2013), 407-415.

Drasner K., (2015). In Katzung B.G. Trevor A.J (eds), Basic

& Clinical Pharmacology (13 ed). New York, NY:

McGraw-Hill.

DuPree, E., & Musheno, D. 2014. A new approach to

preventing falls with injuries. Journal of Nursing

Quality, 29(2), 99-102.

Dykes, P., Hurley, A., Benoit, A., & Middleton, B. (2009).

Why do patients in acute care hospitals fall? Can

falls be prevented? The Journal of Nursing

Administration, 39(6), 299-304.

Dykes, P., Carroll, D., Hurley, A., Lipsitz, S., Benoit, A.,

Chang, F., Meltzer, S., Tsurikova, R., Zuyov, L., &

Middleton, B. (2010). Fall prevention in acute care

hospitals: a randomized trial. Journal of the

American Medical Association, 304(17), 1912-

1918.

Gastrografin® [package insert]. Princeton, NJ: Bracco

Diagnostics Inc. 2013.

Johnson, D., Greenberg, S., Sathiyakumar, V., Thakore, R.,

Ehrenfeld, J., Obremskey, W. T., & Sethi, K.

(2015). Relationship between the Charlson

Comorbidity Index and cost of treating hip

fractures: implications for bundled payment.

Journal of Orthopaedic Traumatology, 16, 209-

213. doi: 10.1007/s10195-015-0337-z.

Krauss, M., Evanoff, B., Hitcho, E., Ngugi, K., Dunagan, W.,

Fischer, I., & Fraser, V. (2005). A case‐control study of patient, medication, and care‐related risk factors for inpatient falls. Journal of General

Internal Medicine, 20(2), 116-122.

Hwang, K., & Kim, G. (2010). Thiazide-induced hyponatremia.

Electrolyte Blood Press, 8(1), 51-57. doi: 10.5049/E

BP.2010.8.1.51

Lam, C. F., Hsieh, S.Y., Wang, J. H., Pan, H. S., Liu, X. Z.,

Ho, Y. C. …& Chen, T. Y. (2016). Incidence and

characteristic analysis of in-hospital falls after

anesthesia. Perioperative Medicine, 5(11), [Online].

doi: 10.1186/s13741-016-0038-z.

Liamis G., Filippatos, D.T., & Elisaf M.S. (2016). Thiazide-

associated hyponatremia in the elderly: what the

clinician needs to know. Journal of Geriatric

Cardiology, 13(2): 175-182.

National Database of Nursing Quality Indicators. (2016).

Guidelines for data collection and submission on

patient falls indicator - inpatient and ambulatory

areas. Overland Park, KS: Press Ganey

Associates, Inc.

Quan, H., Sundararajan, V., Halfon, P., Fong, A., Burnand,

B., Luthi, J.C., & Ghali, W. A. (2005). Coding

algorithms for defining comorbidities in ICD-9-

CM and ICD-10 administrative data. Medical

Care, 43(11), 1130-1139. Retrieved from

http://www.ncbi.nlm.nih.gov/pubmed/16224307.

Quan H., Bing, L., Couris, C., Fushimi, K., Graham, P.,

Hider, P. & Sundararajan, V. (2011). Updating and

validating the Charlson Comorbidity Index and

score for risk adjustment in hospital discharge

abstracts using data from 6 countries. American

Journal of Epidemiology, 173, 676 – 682.

Quigley, P. & White, S. (2013). Hospital-based fall program

measurement and improvement in high reliability

organizations. The Online Journal of Issues in

Nursing, 18(2), manuscript 5.

R Core Team (2015). R: A language and environment for

statistical computing (Version 3.2.1) [Computer

software]. R Foundation for Statistical Computing,

Vienna, Austria. Retrieved from http://www.R-

project.org.

RStudio (2013). RStudio: Integrated development

environment for R (Version 0.97.551) [Computer

software]. RStudio, Inc., Boston, MA. Retrieved

from http://www.rstudio.com

Sardar, G. K., & Eilbert, W. P. (2015). Severe hyponatremia

associated with thiazide diuretic use. Journal of

Emergency Medicine, 48(3), 305-309. doi: 10.101

6/j.jemermed.2014.09.054

Salix Pharmaceuticals. (2016). Relistor® (methylnaltrexone

bromide) – highlights of prescribing information.

Raleigh, NC. Retrieved from https://shared.salix.

com/shared/pi/relistor-pi.pdf?id=8251081.

Schmolders, J., Friedrich, M. J., Michel, R., Strauss, A.C.,

Wimmer, M. D., Randau, T. M.,…& Gravius, S.

(2015). Validation of the Charlson comorbidity

index in patients undergoing revision total hip

arthroplasty. International Orthopaedics, 39(9): 1771-

7.

Slatkin, N., Thomas, J., Lipman, A. G., Wilson, G.,

Boatwright, M. L., Wellman, C., & Israel, R.

(2009). Methylnaltrexone for treatment of opioid-

induced constipation in advanced illness patients.

Journal of Supportive Oncology, 7(1), 39-46.

PubMed PMID: 19278178.

Thomas, J., Karver, S., Cooney, G. A., Chamberlain, B. H.,

Watt, C. K., Slatkin, N. E.,…& Israel, R. J. (2008).

Methylnaltrexone for treatment of opioid-induced

constipation in advanced illness. New England

Journal of Medicine, 358, 2332-2343. Retrieved

from http://www.nejm.org/doi/pdf/10.1056/NEJM

oa0707377.

Tinetti, M., Doucette, J., Claus, E., & Marottoli, R. (1995).

Risk factors for serious injury during falls by older

persons in the community. Journal of the American

Geriatrics Society, 43(11), 12-14-1221.

Tzeng, H. M., & Yin, C. Y. (2008). The extrinsic risk factors

for inpatient falls in hospital patient rooms. Journal

of Nursing Care Quality, 23(3), 233-241.

17

David et al.: An Exploration of the Association of Patient Characteristics and

Published by Scholarly Commons @ Baptist Health South Florida, 2018

U.S. Census Bureau. (2010). Quick facts: Homestead,

Florida [Data file]. Retrieved from https://www.ce

nsus.gov/quickfacts/table/PST045215/1232275,12

22975,1258975,1247700,1239950.

Williams, T., Szekendi, M., & Thomas, S. (2014). An

analysis of patient falls and fall prevention

programs across academic medical centers. Journal

of Nursing Care Quality, 29(1), 19-29.

18

Nursing & Health Sciences Research Journal, Vol. 1, Iss. 1 [2018], Pg. 7-18

https://scholarlycommons.baptisthealth.net/nhsrj/vol1/iss1/4