Potential conflict of interest: Nothing to report.
This study was supported by grant 5348/08 from Col.legi Oficial d'Infermeria de Barcelona, Spain. CIBERehd is supported by the Instituto de Salud Carlos III, Ministerio de Ciencia e Innovación, Spain.
This Article first published online ahead of print on 04 April 2012. The data on page 1927 has since been updated: “These data confirm our previous retrospective study, in which patients with cirrhosis and PHES < −4 reported a higher incidence of falls during the previous year than patients with PHES > or equal to −4 and controls.”
Falls are frequent among patients with debilitating disorders and can have a serious effect on health status. Mild cognitive disturbances associated with cirrhosis may increase the risk for falls. Identifying subjects at risk may allow the implementation of preventive measures. Our aim was to assess the predictive value of the Psychometric Hepatic Encephalopathy Score (PHES) in identifying patients likely to sustain falls. One hundred and twenty-two outpatients with cirrhosis were assessed using the PHES and were followed at specified intervals. One third of them exhibited cognitive dysfunction (CD) according to the PHES (<−4). Seventeen of the forty-two patients (40.4%) with CD had at least one fall during follow-up. In comparison, only 5 of 80 (6.2%) without CD had falls (P < 0.001). Fractures occurred in 4 patients (9.5%) with CD, but in no patients without CD (P = 0.01). Patients with CD needed more healthcare (23.8% versus 2.5%; P < 0.001), more emergency room care (14.2% versus 2.5%; P = 0.02), and more hospitalization (9.5% versus 0%; P = 0.01) as a result of falls than patients without CD. Patients taking psychoactive treatment (n = 21) had a higher frequency of falls, and this was related to an abnormal PHES. In patients without psychoactive treatment (n = 101), the incidence of falls was 32.4% in patients with CD versus 7.5% in those without CD (P = 0.003). In the multivariate analysis, CD was the only independent predictive factor of falls (odds ratio, 10.2; 95% confidence interval, 3.4-30.4; P < 0.001). The 1-year probability of falling was 52.3% in patients with CD and 6.5% in those without (P < 0.001). Conclusion: An abnormal PHES identifies patients with cirrhosis who are at risk for falls. This psychometric test may be useful to promote awareness of falls and identify patients who may benefit from preventive strategies. (HEPATOLOGY 2012;55:1922–1930)
Cognitive dysfunction (CD) is frequent in patients with cirrhosis and without signs of overt hepatic encephalopathy (HE).1–6 The causes of CD can be the result of multiple issues, including the etiology of cirrhosis (e.g., alcohol and hepatitis C), malnutrition (e.g., vitamin deficiencies), sequels of previous overt HE, or other comorbidities (e.g., small vessel cerebrovascular disease secondary to diabetes mellitus or arterial hypertension or psychoactive treatments).1, 3, 5 CD attributable to liver failure and portal-systemic shunting is known as minimal HE (MHE).1, 5, 7 Diagnosis is usually based on the presence of CD with a pattern of subcortical disturbance on psychometric testing (e.g., attention impairment and psychomotor slowing) or neurophysiological abnormalities without alternative causes.1-6
It can be difficult to appraise the relative contribution of comorbidities and MHE on CD without excluding patients with comorbidities. However, such exclusion may decrease representativity in daily clinical practice. For this reason, although we believe that CD, in most patients in our study, mainly corresponds to MHE, in the absence of well-established criteria,5, 7 CD in cirrhosis is more appropriate to describe our population and we preferred therefore to use this term in our study.
CD in cirrhosis has become more relevant in recent years because it has been associated with overt HE,2 mortality,1, 8 worsening in quality of life, and deterioration in daily functioning.1, 3, 9 MHE has a negative effect on driving, and these patients are more predisposed to traffic accidents and violations.10, 11 Because CD impairs attention and reaction capability,1, 4 it likely also predisposes patients with cirrhosis to fall, as we observed in a retrospective assessment.12 However, this association has not yet been prospectively evaluated.
Falls are particularly important in patients with cirrhosis because their risk of fracture is higher than that in the general population.13 This risk has been attributed to a decrease in bone mass resulting from malnutrition, hypogonadism, and liver insufficiency,14 but it could also be a consequence of CD-related falls.12 Moreover, traumas in patients with cirrhosis are a significant cause of complications and mortality.15 In addition to the negative consequences for the patient, falls and fractures have implications for the patient's relatives and are an economic and social burden for the community.16
The Psychometric Hepatic Encephalopathy Score (PHES) consists of a battery of five paper-pencil tests specifically developed for the diagnosis of MHE.2, 4 PHES is scored from the comparison with nomograms in healthy controls; each negative point represents one standard deviation (1 SD) below the mean of the reference population. A result on the PHES <−4 has been proposed for the diagnosis of MHE.2, 4
We designed this study to assess whether, in addition to detecting a cognitive disturbance, the PHES could identify those patients with a higher risk for falls.
AUROC, area under the receiver operating characteristics curve; BMI, body mass index; CD, cognitive dysfunction; CFF, critical flicker frequency; HE, hepatic encephalopathy; MAP, mean arterial pressure; MELD: model for end-stage liver disease score; MHE, minimal hepatic encephalopathy; PHES, Psychometric Hepatic Encephalopathy Score; SD, standard deviation; SSRIs, selective serotonin-reuptake inhibitors; TIPS, transjugular intrahepatic portosystemic shunt; TUG, Timed Up-and-Go Test.
Patients and Methods
One hundred and thirty consecutive outpatients with cirrhosis who visited at two tertiary care hospitals (Hospital de la Santa Creu i Sant Pau and Hospital Vall d'Hebron, Barcelona, Spain) were included from March 2008 to January 2010.
Cirrhosis was diagnosed by clinical, analytical, and ultrasonographic findings or by liver biopsy. Exclusion criteria were as follows: any hospitalization in the previous month resulting from decompensation of cirrhosis, hepatocellular carcinoma, active alcohol intake (in the previous 3 months), current overt acute or chronic HE, cognitive impairment (mini-mental Lobo test <24), neurological disease, inability to perform psychometric tests, marked symptomatic comorbidities (e.g., cardiac, pulmonary, renal, or untreated active depression), or life expectancy less than 6 months. Patients with a follow-up of less than 1 month were excluded from the analysis of the results.
We recorded demographic parameters and clinical and analytical data, such as etiology of cirrhosis, previous decompensations, previous transjugular intrahepatic portosystemic shunt (TIPS), Child-Pugh score, and model for end-stage liver disease (MELD) score.
We also recorded parameters that influence the predisposition to fall in populations other than patients with cirrhosis. These parameters included serum sodium,17 mean arterial pressure (MAP),17, 18 pharmacologic treatment,17-19 body mass index (BMI),18, 19 previous falls,18, 19 degree of comorbidity17-19 determined by the modified Charlson scale,20 visual acuity assessed by Snellen's test,21 and walking problems.17
The PHES includes a neuropsychological battery composed of five different paper-pencil tests: Number Connection Test A and B; Line Tracing Test; Serial Dotting Test; and Digit Symbol Test.4 This battery detects changes in attention and psychomotor speed, which are the areas most affected by HE. We used the computer program of the Red Española de Encefalopatía Hepática (available at: www.redeh.org). The PHES has been validated for the Spanish population, and results were adjusted for age and educational level.22 Patients were considered to have CD when the PHES score was <−4 points.2, 4, 22
Critical flicker frequency
Critical flicker frequency (CFF) is a computerized test to detect MHE in patients with cirrhosis. In our study, CFF was performed as a complementary test. A portable, battery-powered analyzer (Hepatonorm Analyzer; R&R Medi-Business Freiburg GmbH, Freiburg, Germany) was used. In this method, an intermittent red light gradually decreases the initially high-frequency pulse (60 Hz), and when the patient perceives that the light turns from steady to flickering, the frequency at which the patient perceives this change is recorded as the CFF value.2 The procedure was repeated five times to ensure patient understanding. The test was then repeated 10 times, and mean ± SD values were calculated for each patient. CFF was measured in a quiet, semidarkened room to avoid interferences. CFF was not performed in patients with visual defects that precluded accurate performance of the procedure.
After the PHES and CFF tests were performed, falls were assessed by phone interviews with patients every 3 months during a 1-year follow-up.23–25 The researcher who performed the interview was blinded to the presence or not of CD. Interviews were specifically addressed to determine the incidence and characteristics of falls based on a previously described questionnaire.19 Patients' medical records were revised to check and complete the information given by patients and relatives.
To define falls, we used the World Health Organization definition as follows: “A fall is an event which results in a person coming to rest inadvertently on the ground or floor or other lower level.”26 The incidence of falls and the mean number of falls per patient were determined. Severity of injuries and the healthcare needed for falls were also recorded. Fall injuries were classified as contusion, wound, or fracture.12, 27 Healthcare needed was classified as primary care, emergency room care, or hospitalization.12, 28 We also recorded the duration of hospitalization resulting from falls and whether or not patients presented with decompensation of cirrhosis during this admission. Falls were analyzed by comparing patients with cirrhosis and with CD to those without CD, and we evaluated the characteristics of patients according to whether or not they presented with falls during the follow-up.
Timed Up-and-Go Test and Orthostatic Hypotension.
The last 31 patients included in the study completed the Timed Up-and-Go Test (TUG) and were evaluated for the presence of orthostatic hypotension immediately after the PHES and CFF tests were performed. The TUG can be used to assess the risk of falls.29 The test determines the time needed to get up from a chair, walk 3 meters, turn around, and walk back to sit down again without support and in a standardized environment.29 To assess orthostatic hypotension, blood pressure was measured twice before this test: first with the patient seated and then after standing up. Orthostatic hypotension was defined as a decrease in systolic blood pressure of at least 20 mmHg or a decrease in diastolic blood pressure of at least 10 mmHg within 3 minutes of standing.30
Patients with CD were compared with those without CD and patients with falls were compared with those without falls, using Fisher's exact test for categorical variables and the Student's t test and Mann-Whitney's U test for quantitative variables. Parameters that reached statistical significance in the univariate analysis were included in a multivariate analysis by logistic regression to identify the independent factors associated with falls. We used a forward stepwise selection procedure with Wald's test to determine the best model. The predictive ability of the resulting model was evaluated using the area under the receiver operating characteristics curve (AUROC). Probability curves were obtained by the Kaplan-Meier's method and were compared using the log-rank test. Correlations were assessed by Spearman's test. Results are presented as mean ± SD or frequencies. Calculations were performed with the SPSS Statistical Package (version 18.0, 2006; SPSS, Inc., Chicago, IL). A P value of <0.05 was considered statistically significant.
The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the research ethics committee. All patients gave written consent to be included in the study after receiving appropriate information.
Eight of the one hundred and thirty outpatients with cirrhosis initially included in the study were excluded, because follow-up was less than 1 month. The final study sample thus included 122 patients. Mean age in the whole series at inclusion in the study was 63.0 ± 10.1 years, 77 of 122 patients (63.1%) were male, the etiology of cirrhosis was alcohol in 68 of 122 (55.7%), mean Child-Pugh score was 6.2 ± 1.5, 106 of 122 (86.8%) had decompensated cirrhosis, 26 of 122 (21.3%) presented with mild ascites, and no patients showed signs of HE. Thirty-one of one hundred and twenty-two patients (25.4%) had a previous episode of overt HE, and 21 of 122 (17.2%) were taking psychoactive treatment.
Forty-two of the one hundred and twenty-two patients (34.4%) had CD (i.e., PHES <−4), and 80 (65.5%) did not. Table 1 shows the characteristics of patients in each group. Patients with CD were older and they were more frequently women. They had more advanced liver insufficiency, more previous episodes of HE, more previous TIPS, lower levels of hemoglobin and serum sodium, more previous falls, and a higher incidence of diabetes. Furthermore, in this group, more patients had ascites and more were taking nonabsorbable disaccharides and antidepressants. CFF was performed in 93 patients (28 with CD and 65 without CD) and was more impaired in the former.
Table 1. Clinical and Analytical Characteristics of Patients With and Without CD (PHES <−4)
CD (n = 42)
No CD (n = 80)
Results are expressed as number of patients (%) or mean ± SD.
Number of patients taking sedatives and/or antidepressants. Five patients with CD and 3 patients without CD were taking both sedatives and antidepressants.
In the subgroup of patients that completed the TUG, those with CD (n = 12) took longer to perform the test than those without CD (n = 19) (15.2 ± 4.5 versus 12.3 ± 2.6 seconds; P = 0.06). Orthostatic hypotension was present in 1 of 12 (8.3%) patients with CD and in 2 of 19 (10.5%) patients without CD (P = 1.0).
Mean follow-up was 9.5 ± 3.7 months in patients with CD and 11.2 ± 2.1 in patients without CD (P = 0.008). This difference was the result of a higher mortality in patients with CD (10 of 42 [23.8%] versus 3 of 80 [3.8%]; P = 0.001). The incidence of falls during the follow-up is shown in Fig. 1. The percentage of patients who presented at least one fall during follow-up was higher in patients with CD (17 of 42; 40.4%) than in patients without CD (5 of 80; 6.2%) (P < 0.001). The total number of falls was 32 in patients with CD and 6 in those without CD. The mean number of falls per patient was therefore higher in CD patients than in non-CD patients (0.76 ± 0.21 versus 0.08 ± 0.30; P = 0.003). No patient fell during an episode of overt HE.
Figure 2A shows the type of injuries resulting from falls. Patients with CD had a higher incidence of both mild injuries (e.g., contusions) and severe injuries (e.g., fractures) than those without CD. In patients with CD, there were 23 contusions in 14 patients, 3 wounds in 2 patients, and 6 fractures in 4 patients. In patients without CD, there were 5 contusions in 4 patients, 1 wound in 1 patient, and no fractures. The 6 fractures in CD patients were 1 Colles, 1 femur, 1 costal, 1 vertebral, and 2 humerus fractures. Surgery was only required for the patient with the femur fracture.
Figure 2B shows healthcare requirements as a consequence of falls. Patients with CD required healthcare resulting from falls more often than patients without CD (10 of 42 [23.8%] versus 2 of 80 [2.5%]; P < 0.001). Considering the type of medical care needed, patients with CD required more emergency room care and more hospitalizations than patients without CD. In patients with CD, 10 patients needed healthcare resulting from falls in 12 episodes of fall: 1 patient needed primary care in 1 episode, 6 patients needed emergency room care in 6 episodes, and 4 patients needed hospitalizations in 5 episodes. In patients without CD, 2 patients needed healthcare resulting from falls in 2 episodes of fall, both attended in the emergency room. If a patient had several types of healthcare requirements as a result of one episode of fall, we considered only the most complex.
Two of the four patients with CD who were hospitalized as a result of falls developed decompensation of cirrhosis during admission (1 encephalopathy and 1 ascites). In contrast, no patient without CD required hospitalization as a result of falls. The total number of days of hospitalization as a result of falls was 54 for the 42 patients with CD and 0 for the 80 patients without CD. The mean number of days of hospitalization as a result of falls per patient was 1.29 ± 4.6 in patients with CD versus 0 ± 0 in patients without CD (P = 0.08).
Table 2 shows the relationship between falls and CD, stratified by age, gender, treatment with psychoactive drugs (e.g., antidepressants and/or sedatives), compensated versus decompensated cirrhosis, and previous HE. Patients who presented with CD fell more than those without CD, considering patients on psychoactive treatment and also patients not taking these drugs. When analyzing only patients with CD, patients taking psychoactive drugs fell more than patients who were not taking these treatments. In patients younger than 65 years old and patients without previous overt HE, the incidence of falls was also significantly higher in patients with CD than in those without.
Table 2. Incidence of Falls in Patients With and Without CD (PHES <−4) Stratified by Age, Gender, Psychoactive Treatment, Compensated versus Decompensated Cirrhosis, and Previous Encephalopathy
CD (n = 42)
No CD (n = 80)
Results are expressed as number of patients (%).
P = 0.04 with respect to patients with CD and no psychoactive drugs.
P = 0.04 with respect to patients with CD and no antidepressants.
P value was nonsignificant in the remaining comparisons of stratification groups, both in patients with CD and in patients without CD.
Table 3 shows the characteristics of patients who fell and patients who did not. In the univariate analysis, among patients who fell during follow-up, there were more women, CD was more frequent, the PHES score was lower, and more patients took antidepressant treatment than in patients who did not fall.
Table 3. Clinical and Analytical Data of Patients According to Whether They Presented Falls or Not
Falls (n = 22)
No Falls (n = 100)
Results are expressed as number of patients (%) or mean ± SD.
Number of patients taking sedatives and/or antidepressants. Four patients with falls and 4 patients without falls were taking both sedatives and antidepressants.
Multivariate analysis, including gender, antidepressant treatment, and cognitive dysfunction, showed that CD (odds ratio, 10.2; 95% confidence interval, 3.4-30.4; P < 0.001) was the only independent factor associated with falls during follow-up; the area under the receiver operating characteristics curve (AUROC) was 0.76 (P < 0.001). Figure 3 shows that the 1-year probability of falling was higher in patients with CD (52.3%) than in patients without (6.5%) (P < 0.001).
In the subgroup of patients that completed the TUG, the test took longer to complete in patients with falls (n = 11) than in patients without falls (n = 20) (15.6 ± 4.4 versus 12.3 ± 2.6 seconds; P = 0.03). Orthostatic hypotension was present in 0 of 11 patients who fell and in 3 of 20 (15%) who did not fall (P = 0.53).
Figure 4 shows the total number of patients and the number of patients who fell for each PHES value. Falls began to occur especially at −5 points, but the incidence did not increase in parallel with worsening of the PHES score. Moreover, considering patients with CD (i.e., PHES <−4), there was no correlation between PHES score and the number of falls (r = −0.08; P = 0.60), and PHES score was similar in patients who fell (n = 17) and in those who did not fall (n = 25) (−7.1 ± 2.0 versus −7.0 ± 1.7; P = 0.76).
This is the first prospective study showing that CD defined by an impaired PHES is a predictive factor of falls in outpatients with cirrhosis. Patients with CD had an incidence of falls of 40.4%, in contrast with patients without CD who had an incidence of 6.2%. Moreover, the probability of CD patients falling was 52.3% at 1-year follow-up. These data confirm our previous retrospective study, in which patients with cirrhosis and PHES <−4 reported a higher incidence of falls during the previous year than patients with PHES > or equal to −4 and controls.12
In agreement with previous data, in the present study, one third of patients had CD,1-3 and this condition was associated with factors such as age,31 the degree of liver insufficiency,32 previous episodes of overt HE,32, 33 TIPS,34 hyponatremia,35 or CFF results.2
In populations other than patients with cirrhosis, mainly in elderly people and patients with neurological diseases, predisposition to falling has been related to a wide range of factors, including age, gender, previous falls, hyponatremia, hypotension, pharmacological treatment, degree of comorbidity, impairment in visual acuity, walking problems, or BMI.17-19, 21 We assessed most of these factors in the present study, and comparing patients who fell to patients who did not, we found statistical differences only in gender, antidepressant treatment, and CD assessed by the PHES. However, the multivariate analysis identified CD as the only independent predictive factor of falls. Interestingly, the higher incidence of falls in patients with PHES <−4 was also evident when we analyzed only patients below 65 years old, patients without psychoactive treatment, and patients without previous overt encephalopathy. These data strongly support the association of CD with falls in patients with cirrhosis.
A clear relationship between treatment with selective serotonin-reuptake inhibitors (SSRIs) and falls and fractures has been described in the general population.17-19 In our study, antidepressant treatment was a factor associated with CD and also with falls. This association might have been favored by the effects of SSRIs on serotonin metabolism36 and the impaired hepatic clearance of these drugs37 in the setting of cirrhosis. Because patients with CD taking psychoactive medication showed the highest incidence of falls, we hypothesize that CD related to cirrhosis and treatment with psychoactive drugs may have a cumulative effect on predisposition to falling.
In the present study, the incidence of falls was higher in women than in men. This gender difference has also been observed in the general population18, 19, 38 and is thought to be related to lower muscle strength and speed of muscle contraction in women.39 Moreover, in our study, CD was more frequent in women than in men. This could also have contributed to this finding.
The precise mechanisms by which an impaired PHES is associated with falls are not known. They could be related to cognitive impairment in cirrhosis, mainly affecting attention, visuomotor coordination, psychomotor speed, and reaction times.1, 4, 6, 12 Such a relationship between cognitive impairment and falls has been observed in elderly patients17 and in stroke survivors.40 However, in our study, there was no relationship between incidence and number of falls per patient and severity of PHES impairment when considering only patients with CD according to PHES ≤4. Moreover, CFF was not statistically different between patients who fell and those who did not. CFF mainly measures attention and reaction capability.2, 34 These findings suggest that the main cause for predisposition to falling is not CD assessed by the PHES, but a coincident neuromuscular disturbance.
One possibility is that the higher incidence of falls in patients with altered PHES might be related to parkinsonism associated with cirrhosis.41, 42 Parkinsonism in patients with cirrhosis is frequent and related to cognitive impairment and worsening in daily-life activities.41 In the present study, extrapyramidal signs were not specifically assessed. However, we evaluated the TUG in a subgroup of patients, and those with falls took longer to perform the test. This tool is used to assess the risk of falls, and scores are higher when gait and balance disorders are present,29 as in patients with Parkinson's disease.43 This finding supports the possible role of parkinsonism in the predisposition of patients with CD to fall.
Falls in patients with cirrhosis could also be the result of decreased muscle strength.18, 44 Although muscular function was not evaluated in the present study, muscle weakness is frequent in patients with cirrhosis and has been associated with cognitive impairment.45 A recent retrospective study has shown that patients with primary biliary cirrhosis in the noncirrhotic stage fell more than controls, and falling was associated with impairment in lower limb strength.46
When we analyzed the severity of injuries and the healthcare requirements resulting from falls, we observed that the incidence of severe injuries and the need for hospital care were both significantly higher in patients with CD than in those without. In patients with CD, 9.5% sustained a fracture and one quarter of them required hospital care as a consequence of a fall. These data suggest that patients with CD are prone not only to falls resulting in mild injuries, but also, at least in the same proportion, to severe injuries. This also avoids, in part, the bias that could result from the method used to assess the incidence of falls in the present study. The method of periodic interview is widely used to evaluate the occurrence of falls in populations other than patients with cirrhosis,23-25 but some falls could be missed if the patients do not remember them when the interview is administered.23, 28 This bias is minimized when we assess falls that cause severe injuries because these are more difficult for the patients and their relatives to forget. Furthermore, they are recorded on the clinical records.
Considering the importance of falls in patients with cirrhosis and CD, it seems reasonable to develop strategies addressed at their prevention. These strategies should include promoting the use of the PHES to identify patients at risk of falls, consider treating CD with antibiotics, such as rifaximin,47 nonabsorbable disaccharides,48 or probiotics,49 and also the rational use of psychoactive drugs.18, 19 In the general population, multifactorial interventions (including recommendations for precautions in daily life), exercise to increase muscle strength and balance, and vitamin D supplements have proven useful in preventing falls.50 These measures could also be helpful in cognitive-impaired patients with cirrhosis.
We conclude that CD identified by an impaired PHES is a factor associated with falls in patients with cirrhosis. Falls in these patients are a significant cause of morbidity and healthcare requirements. Further studies are warranted to address the mechanisms implicated in this predisposition and to design preventive strategies.
The authors thank Carolyn Newey for her English language revision.