Professional Documents
Culture Documents
Assessing Functional Status After Intensive Care Unit Stay - The Barthel Index and The Katz Index
Assessing Functional Status After Intensive Care Unit Stay - The Barthel Index and The Katz Index
doi: 10.1093/intqhc/mzx203
Advance Access Publication Date: 27 January 2018
Article
Article
Address reprint requests to: Carolina Fu, Department of Physiotherapy, Communication Science & Disorders,
Occupational Therapy – Medical School of University of São Paulo, Rua Cipotânea 51, Cidade Universitária, 05360-000
São Paulo-SP, Brazil. Tel: +55-11-30918421; Fax: +55-11-30917459; E-mail: carolfu@usp.br
Editorial Decision 6 December 2017; Accepted 9 January 2018
Abstract
Objective: To assess the functional status of post-ICU patients using the Barthel Index (BI) and the
Katz Index (KI) and to assess which is more suitable for this population.
Design: Retrospective longitudinal study.
Setting: Public tertiary hospital in São Paulo (Brazil).
Participants: Patients aged ≥18 years old, admitted to ICU, who were treated with mechanical ven-
tilation (MV) ≥ 24 h and were discharged to ward.
Exclusion criteria: Inability to answer the BI and the KI; limiting neurological or orthopaedic condi-
tions; ICU stay ≥90 days. Patients transferred to or from other hospitals or who died in the wards
were not analysed.
Intervention: BI and KI were scored pre-ICU and post-ICU and the variation was calculated.
Main Outcome Measures: BI and KI scores were compared using analysis based on item response
theory (IRT), using degree of difficulty and discriminating items as parameters.
Results: Median age was 52 years old, median APACHE II score was 15. Median ICU stay was 11
days and median MV duration was 4 days. BI variation was 44% and KI variation was 55%. In IRT
analysis, BI considered a larger number of items with different levels of difficulty.
Conclusion: Both the BI and the KI revealed significant deterioration of functional status after ICU
discharge. The IRT analysis suggested that the Barthel Index might be a better scale than the Katz
Index for the assessment of functional status of patients discharged from ICU, since it presented
better discrimination of the ability to carry out the tasks.
Key words: intensive care, activities of daily living, recovery of function, outcome assessment
© The Author(s) 2018. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com 265
Relative variation for Katx index Table 1 Clinical and demographic data
Post ICU score − Pre ICU score
= N = 249
18 − Pre ICU score
Age, years 52 (36–61)
Since some of the tasks on the BI and KI had more than two possible
BMI, kg/cm2 23.9 (23.8-23.7)
answers, the responses for each task of the BI and KI were categorized.
Gender, male, n (%) 133 (53)
Thus, patients were categorized into ‘able to do,’ if they were totally APACHE II score on ICU admission day 15 (11–21)
independent in perform the task, and ‘not able to do,’ if they presented Charlson index adjusted for age 3 (1–5)
any level of dependence or were unable to perform the task at all. Need for renal replacement therapy, n (%) 162 (65)
Additional analysis of the BI and KI pre-ICU and post-ICU was Type of hospital admission, n (%)
carried out, using item response theory (IRT) [14, 15] to assess Non-surgical 94 (38)
whether one of the indexes was preferable for post-ICU patients. Emergency surgery 71 (28)
Analysis based on IRT is a statistical technique that allows identify- Elective surgery 42 (17)
ing which items present a higher degree of difficulty, and which Clinical emergency 42 (17)
Cause of hospital admission
items are able to discriminate the investigated population. The IRT
Respiratory 40 (16)
model included two parameters – difficulty and discrimination –
Gastrointestinal 49 (20)
considering the dichotomous answers ‘able to do’ and ‘not able to Pre-surgery 42 (17)
do,’ as mentioned above. The latent variable was functional status. Trauma 23 (9)
The characteristics of each scale were verified using visual Cardiovascular 20 (8)
graphic analysis in IRT. The flatter the curve, the more discrimin- Infection 18 (7)
atory the item. The projection of the central point of the curve (loca- Neurological 14 (6)
tion parameter, that represents the latent score value needed to get a Clinical investigation 10 (4)
50% probability of being able to do the item) reflects the difficulty Others* 33 (13)
of the item. Item characteristic curves show how the probability of Cause of ICU admission, n (%)
Post-operative** 106 (43)
responding in each item changes with the values of the latent vari-
Respiratory insufficiency 64 (26)
able (ability), as well as with the difficulty parameters of the items.
Shock 40 (16)
For curves with positive slope (positive discrimination parameter), Coma 17 (7)
when the projection of the central point of the curve is to the right Renal insufficiency 11 (4)
of the graph, the difficulty of the task is higher. Statistical analysis Monitoring*** 5 (2)
was performed with the statistical software R, version 3.1.3 [16]. Cardiorespiratory arrest 3 (1)
Liver insufficiency 3 (1)
BMI: body mass index; APACHE II: Acute Physiology and Chronic Health
Results
Evaluation; ICU: intensive care unit. Data are expressed as n (%), where indi-
A total of 858 patients were assessed for eligibility. Excluded cated, or median (1st−3rd interquartile). *Others: refers to other grouped
patients were 183 patients with cognitive impairment, 102 patients causes of hospital admission and includes: renal, metabolic, haematological,
with neurologic disease after hospital admission, 71 patients with exogenous intoxication, bleeding, skin disease, tetanus. **Surgery for patients
aphasia or communication impairment, 63 patients were trans- admitted to the ICU in post-operative routine were the following: abdominal
or gastrointestinal surgery (39 patients, 37%); vascular surgery (24 patients,
ferred, died or were discharged <48 h after ICU discharge, 40
23%); liver transplantation (21 patients, 20%), head and neck surgery
patients could not be assessed within 48 h after ICU discharge, 53
(11 patients, 10%); urogynecology surgery (4 patients, 4%); neurological sur-
patients presented lower limb amputation or bed rest prescription,
gery (3 patients, 3%); soft tissue surgery (2 patients, 2%); thoracic surgery
44 patients presented previous neurologic or neuromuscular disease, (1 patient, 1%) and kidney transplantation surgery (1 patient, 1%).
8 patients presented ICU stay >90 days, 24 patients refused to par- ***Monitoring refers to patients admitted to the ICU with clinical deterior-
ticipate, 9 patients were transferred to another hospital, 7 patients ation, but without known diagnosis.
died and 5 patients were discharged to long-term institutes.
Therefore, 249 patients were included in the study.
Tables 1 and 2 present the population characteristics. The Before ICU admission, all items of both questionnaires presented
median age was 52 years old, so this was a non-elderly population, the same level of difficulty for most of the tasks, since the curves pre-
and median APACHE II score within the first 24 h of ICU admission sent the same position – except for the item ‘feeding.’ This item is
was 15 points. Comorbidity, evaluated by the Charlson index located in a position to the left of the other curves, indicating that
adjusted by age, presented a median of 3 points. The main cause of the probability of being able to do this task is lower for those
ICU admission was post-surgery. The median ICU length of stay patients who present lower scores according to the IRT scale.
was 11 days and the median MV duration was 4 days. After ICU discharge, the curves became more flattened than at
Table 3 presents total scores on the BI and KI. Patients suffered ICU admission. This indicates that the patients are now less homo-
deterioration of functional status after ICU stay. Fig. 1 provides geneous regarding the ability to perform the tasks. Also, the projec-
more detail about this observation. Before ICU admission, the pro- tion of the central point of the curves changes, showing that the
portion of patients who were unable to do any of the tasks on the BI items take on different levels of difficulty. This was observed in both
and KI was no more than 3.2%. After ICU discharge, the smallest questionnaires, but mostly for the BI. This suggests that after ICU
percentage of patients who were unable to perform a task was discharge, the BI considers a larger number of items with different
33.6%. Depending on the task, the percentage of patients who were levels of difficulty than before ICU admission. Thus, it may be a bet-
incapable of doing it was even greater. For example, 88.3% of ter tool for the assessment of the type of population involved in this
patients were not able to climb stairs. study. This is illustrated in Fig. 2.
Figure 1 Proportion of patients ‘able to do’ or ‘not able to do’ each activity of daily living of Barthel Index and Katz Index, considering dichotomous answers.
ICU: Intensive Care Unit.
Figure 2 Characteristic curve of each item for Barthel Index and Katz Index before (above) and after (below) ICU admission. The curve on the left side of the
graphics for before ICU admission represents the task ‘feeding.’ This figure indicates that before ICU admission, if the patient presented a lower Barthel Index
score, they presented lower probability of being able to perform the task ‘feeding.’ All the other curves presented the same level of discrimination and difficulty.
After ICU discharge, the curves become more flattened and the projection of the central point of the curves change, mostly for the Barthel Index. It is possible to
observe that the curves are more scattered for the Barthel Index than for the Katz Index.
90% of the study population for the present research presented and the KI were able to detect this fact. These tools have also
maximum independence scores. However, post-ICU patients pre- been shown to be associated with long-term prognosis in some
sented substantial impairment in functional status. Both the BI patients [17, 23, 24].
In contrast with most studies, in which assessment of functional 5. van der Schaaf M et al. Functional status after intensive care: a challenge
status occurred some time after ICU discharge [1, 5, 6], it was evalu- for rehabilitation professionals to improve outcome. J Rehabil Med 2009;
ated immediately after ICU discharge in the present study. This may 41:1041–8.
6. Hartigan I. A comparative review of the Katz ADL and the Barthel Index
be useful for the identification of patients whose functional status is
in assessing the activities of daily living of older people. Int J Older
most impaired, who thus might need a higher level of assistance
People Nurs 2007;2:204–12.
after ICU discharge.
7. Sangha H, Lipson D, Foley N et al. A comparison of the Barthel Index
The transference of information regarding patients’ status at the and the Functional Independence Measure as outcome measures in stroke
time of ICU discharge ensures that critical information is provided to rehabilitation: patterns of disability scale usage in clinical trials. Int J
the care teams that are going to receive the patients [25]. One study Rehabil Res 2005;28:135–9.
showed that 80% of nurses and doctors included information such as 8. Houlden H, Edwards M, McNeil J et al. Use of the Barthel Index and the
patients’ condition and action plans during handover [26], and func- Functional Independence Measure during early inpatient rehabilitation
tional status information may be helpful for both. Functional status after single incident brain injury. Clin Rehabil 2006;20:153–9.
scales provide quantitative information that can be used for evaluation 9. Collin C, Wade DT, Davies S et al. The Barthel ADL Index: a reliability
study. Int Disabil Stud 1988;10:61–3.
of patients’ evolution during hospital stay. Finally, functional status
10. Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md
assessment can help in the optimization of human resource allocation
State Med J 1965;14:61–5.
and, consequently, improve patient care and safety. It can also help in
11. Katz S, Ford AB, Moskowitz RW et al. Studies of illness in the aged. The
rehabilitation planning in the hospital after ICU discharge. Improving Index of ADL: a standardized measure of biological and psychosocial
rehabilitation may reduce long-term restrictions in ADL [5]. function. JAMA 1963;185:914–9.
Limitations of the present study: the IRT analysis did not take 12. Roedl KJ, Wilson LS, Fine J. A systematic review and comparison of func-
into account different levels of assistance; even so, it was able to tional assessments of community-dwelling elderly patients. J Am Assoc
show different behaviour for the BI and KI and point out which Nurse Pract 2016 Mar;28:160–9.
scale would be more suitable for post-ICU patients. The study only 13. SilvaJMNBarbosaMFSCastroPOCNet al2013Correlation between the
evaluated patients who had been submitted to IMV, which is related risk of falling and functional autonomy in institutionalized elderlyRev
Bras Geriatr Gerontol1633746
to bed rest and, thus, to functional decline. Patients who presented
14. Andrade DF, Tavares HR, Valle RC. Teoria da Resposta ao Item: Conceitos
conditions such as delirium, aphasia, and previous or current neuro-
e Aplicações. São Paulo: Associação Brasileira de Estatística, 2000.
logical disorders were excluded because the focus of the study was
15. Baker FB, Kim S. Item Response Theory – Parameter Estimation
evaluating the impact of the ICU stay itself on the functional status Techniques, 2nd edn. New York: Marcel Dekker, Inc, 2004.
of the patients. Therefore, the results should be limited to the studied 16. R Development Core Team. R: A Language and Environment for Statistical
population. Finally, assessment of the BI and KI previous to hospital Computing. Vienna: R Foundation for Statistical Computing, 2005.
admission was made retrospectively. Although the BI and KI con- 17. Silva JMN, Barbosa MFS. Castro POCN, Noronha MM. Correlation
cern very simple tasks, this could cause information bias. between the risk of falling and functional autonomy in institutionalized
In conclusion, in a tertiary public hospital in a developing coun- elderly. Rev Bras Geriatr Gerontol 2013;16:337–46.
try, patients presented a significant decline in functional status after 18. Monacelli F, Tafuro M, Molfetta L et al. Evaluation of prognostic indices
in elderly hospitalized patients. Geriatr Gerontol Int 2016. doi:10.1111/
ICU discharge. Both the BI and KI were able to detect functional sta-
ggi.12801.
tus impairment, and the item response theory analysis suggested that
19. Ferreira NA, Lopes AJ, Ferreira AS et al. Determination of functional
the BI might be a better scale than the KI for the assessment of func-
prognosis in hospitalized patients following an intensive care admission.
tional status in this situation. World J Crit Care Med 2016;5:219–27.
20. Mazière S, Couturier P, Gavazzi G. Impact of functional status on the
onset of nosocomial infections in an acute care for elders unit. J Nutr
Acknowledgements
Health Aging 2013;17:903–7. doi:10.1007/s12603-013-0370-7.
We thank José Marcelo e Souza Mafra, Michele de Cássia Santos Ramos, Cauê 21. KowalikRSzczerbaEKołtowskiŁet al2014Cardiac arrest survivors treated
Padovani and Bruna Peruzzo Rotta, who all helped with data collection. with or without mild therapeutic hypothermia: performance status and
quality of life assessmentScand J Trauma Resusc Emerg Med2276doi:10.
1186/s13049-014-0076-9
Funding 22. Martínez-Velilla N, Cadore L, Casas-Herrero Á et al. Physical activity
There was no funding for this study. and early rehabilitation in hospitalized elderly medical patients: system-
atic review of Randomized Clinical Trials. J Nutr Health Aging 2016;20:
738–51. doi:10.1007/s12603-016-0683-4.
References 23. Oliveira RP, Rosa R, Ascoli A et al. Factors influencing decline of physical
1. Rydingsward JE, Horkan CM, Mogensen KM et al. Functional status in functional status among icu survivors: a prospective cohort study.
ICU survivors and out of hospital outcomes. Intensive Care Med Exp Intensive Care Med Exp 2015;3:A361.
2015;3:A178. 24. Rodríguez-Villar S, Fernández-Méndez R, Adams G et al. Basal functional
2. Hermans G, Van den Berghe G. Clinical review: intensive care unit status predicts functional recovery in critically ill patients with multiple-
acquired weakness. Crit Care 2015;19:274. organ failure. J Crit Care 2015;30:511–7.
3. Sommers J, Engelbert RH, Dettling-Ihnenfeldt D et al. Physiotherapy in the 25. Halvorson S, Wheeler B, Willis M et al. A multidisciplinary initiative to
intensive care unit: an evidence-based, expert driven, practical statement and standardize intensive care to acute care transitions. Int J Qual Health
rehabilitation recommendations. Clin Rehabil 2015;29:1051–63. Care 2016;28:615–25.
4. Elliott D, Denehy L, Berney S et al. Assessing physical function and activ- 26. Kowitlawakul Y, Leong BS, Lua A et al. Observation of handover process
ity for survivors of a critical illness: a review of instruments. Aust Crit in an intensive care unit (ICU): barriers and quality improvement strategy.
Care 2011;24:155–66. Int J Qual Health Care 2015;27:99–104.