Clinical Nutrition (2003) 22(3): 235–239

r 2003 Elsevier Science Ltd. All rights reserved.

doi:10.1016/S0261-5614(02)00215-7

ORIGINAL ARTICLE

The impact of malnutrition on morbidity, mortality, length of

hospital stay and costs evaluated through a multivariate
model analysis
M. ISABEL T. D. CORREIA,* DAN L.WAITZBERGy
*Department of Surgery, Faculdade de Medicina da, Universidade Federal de Minas Gerais, Brazil, y Department of Gastroenterology 
Digestive Surgery Discipline, Faculdade de Medicina da, Universidade de Sao Paulo, Brazil (Correspondence to: MITDC, Department of
Surgery, Faculdade de Medicina da, Universidade Federal de Minas Geris, Rua Gonc%alves Dias 332 apt. 602 Belo Horizonte, MG 30140-
090, Brazil)

AbstractMalnutrition has been identi¢ed as a¡ecting patient outcome. The purpose of this study was to correlate the
nutritional status of hospitalized patients with their morbidity, mortality, length of hospital stay and costs. The patients
were nutritionally assessed within the ¢rst 72 h of hospital admission. The patients’ charts were surveyed on the inci-
dence of complications and mortality. Hospital costs were calculated based on economic tables used by insurance com-
panies. Multivariate logistic regression analysis and the Cox regression model were used to identify possible
confounding factors. A Po0.05 was considered statistically signi¢cant.
The mean age was 50.6717.3 years with 50.2% being male.The incidence of complications in the malnourished was
27.0% [Relative risk (RR)=1.60]. Mortality in the malnourished patients was 12.4% vs 4.7% in the well nourished (RR =
2.63). Malnourished patients stayed in the hospital for 16.7724.5 days vs10.1711.7 days in the nourished. Hospital costs
in malnourished patients were increased up to 308.9%.
It was concluded that malnutrition, as analyzed by a multivariate logistic regression model, is an independent risk
factor impacting on higher complications and increased mortality, length of hospital stay and costs.
r 2003 Elsevier Science Ltd. All rights reserved.

Key words: malnutrition; morbidity; mortality; length
of hospital stay; hospital costs
Introduction
Hospital malnutrition has been related to increased
morbidity, mortality, length of hospital stay and costs
(1–4). The functional and metabolic body derangements,
which justify the previous events, are based on the
premise that malnutrition interferes in almost every
organ and/or system of the human body. The gut of
malnourished patients presents with impaired immune
function, digestion and absorption (5, 6). Muscle
dysfunction, especially of thoracic muscles might
explain the high incidence of pneumonias in the
malnourished (7). Wound healing is also adversely
affected by malnutrition (8). Thirty to fifty percent of
hospitalized patients may present with malnutrition
(9–12) and, despite its high prevalence, medical aware-
ness of the patients’ nutritional status seems to be
lacking (12, 13).
The patient’s disease alone or, in conjunction with
social segregation, psychological factors, economic
status, lack of medical awareness and longer hospitali-
zations, among others, are considered risk factors for
the development of malnutrition (12). Morbidity,
mortality, length of hospital stay (LOS) and hospital
costs are also impacted by other factors in addition to
malnutrition. Therefore, it would be too simplistic to
analyze all of them using statistical univariate analysis.
To avoid influence of other possible confounding
variables, multiple logistic regression analysis should
be applied (14).
The aim of this study was to correlate, by using a
multivariate logistic model, the nutritional status of
hospitalized patients with the incidence of complications,
mortality and LOS. Hospital costs were also estimated.
Methods
Patients
This was a retrospective cohort study review of 709
adult patients randomly selected from 25 Brazilian
hospitals. They were part of a larger in hospital
malnutrition prevalence study previously published
(12). Hospital administration and ethical committee
consent to participate were obtained. The inclusion
criteria were: age above 18 years old and those patients
nutritionally assessed within the first 72 h post-hospital
admission. The exclusion criteria were: previous hospital

235
236 THE IMPACT OF MALNUTRITION

admission within the last 6 months and obstetric Well nourished

patients.
65,8%
Moderately or
suspected
7.9%
Nutritional assessment malnourished
26.3% 65.8%

The nutritional status of the patients was assessed by the Severely

Subjective Global Assessment (15) technique. Patients malnourished
were evaluated within the initial 72 h of hospitalization.
For the purpose of statistical analyses, patients were Fig. 1 Nutritional status at hospital admission, 709 patients.
divided into two groups: well nourished and malnour-
ished (this included moderately or suspected malnutri- Table 1 Nutritional status vs incidence of complications
tion, and the severely malnourished, as defined by Nutritional status Complications Without
Subjective Global Assessment). complications
Well nourished 78 (16.8%) 385 (83.2%)
Complications, mortality, LOS and hospital costs Malnourished 64 (27.0%)n 173 (73.0%)
(moderately and severely)
The patients’ charts were reviewed searching for Only the severely malnourished 24 (42.8%)y 32 (57.2%)
complications, which were divided into two groups: n
RR=1.60 (IC=1.20–2.14; Po0.01).
y
infectious and non-infectious, according to the clinical RR=2.54 (IC=1.77–3.66; Po0.01).
criteria, biochemical and radiological tests, as previously
published by Buzby et al. (16). Table 2 Incidence of infectious complications vs the nutritional status
The causes of death were considered those registered
Infectious complications Well Malnourished
by the attending physician in the medical chart. LOS nourished (moderately and severely)
was determined from the day of admission to the
Pulmonary infection 19nn 17nn
moment of discharge or death. (4.1%) (5.9%)
Costs were calculated based on the information Urinary infection 9nn 9nn
provided by a Medical Cooperative (Unimed-BH) one (1.9%) (3.7%)
Wound infection 10nn 4nn
of the leading insurance plans, in the country. For the (2.1%) (1.7%)
calculations the daily hospital infirmary rate was used. Sepsis 5n 9n
Costs for drugs, tests and materials were only used to (1.1%) (3.7%)
Intrabdominal abscess 2n 5n
estimate the costs of patients with the most common (0.4%) (2.1%)
infectious complications, which were pulmonary infections. Extraperitoneal abscess 1nn 2nn
(0.2%) (0.8%)
Septic coagulopathy 1nn 1nn
Statistical analysis (0.2%) (0.4%)
Total 47n 47n
Univariate analyses using the chi-square test were done (10.1%) (19.4%)
while, introducing the Yates correction when necessary. n nn
Po0.01; NS.
The relative risks (RR) between the risk factor variable
(presence of malnutrition) and the outcome variables
(morbidity, mortality and length of hospital stay) were gynecological and urological diseases (14.6%) and
calculated with their confidence intervals (CI). Those respiratory problems (9.8%). Only 3% were trauma
variables considered risk factors for complications and patients. Cancer was present in 18.2% of the cases.
mortality (by the univariate analysis) were entered into The overall incidence of complications was 20.3%,
the multivariate logistic regression model, to calculate the but a significantly higher percentage was present in the
odds ratio (OR) and its confidence interval. The Cox malnourished patients as shown in Table 1. The list of
multivariate model was used to assess the relationship infectious and non-infectious complications can be seen
between risk factors and LOS. Statistical significance was in Tables 2 and 3.
set at Po0.05. Data were analyzed using EpiInfo (CDC), Other risk factors, such as the presence of cancer and
version 6.0, and SPSS (SPSS Inc.), version 6. 1. 2. infection, age above 60 years old and those undergoing
clinical treatment, were associated with complications,
by the univariate analysis. Therefore, all of these
Results variables were entered in the multiple logistic regression
model. Those, which were statistically different, can be
A total of 709 patients were studied. The mean age seen in Table 4.
was 50.6717.3 years old with 50.2% being male. The Length of hospital stay was shorter in the well-
prevalence of malnutrition at the time of hospitalization nourished patients (10.1711.7 days, median of 6 days vs
can be seen in Fig. 1. The most important causes of 16.7724.5 days, median of 9 days), as can be seen in
hospitalization were cardiovascular diseases in 20.8% of Figures 2 and 3. Other variables were also considered by
the cases, followed by gastrointestinal diseases (18.9%), the univariate analysis, such as protective factors
 CLINICAL NUTRITION 237

Table 3 Incidence of non-infectious complications vs nutritional 400

status
Non-infectious Well Malnourished
complications nourished (moderately and severely)
Respiratory failure 6n 15n 300

Number of patients
(1.3%) (6.2%)
Cardiac arrest 7n 14n
(1.5%) (5.8%)
Others 8nn 9nn 200
(1.7%) (3.7%)
Cardiac arrythmia 7nn 5nn
(1.5%) (2.1%)
Cardiac failure 3n 6n
(0.6%) (2.5%) 100
Wound dehiscence 8nn 1nn
(1.7%) (0.4%)
Total 39n 50n
(8.4%) (20.5%)
0
n
Po0.01; nn
NS. 0 50 100 150 200 250 300 350
Length of hospital stay in days

Table 4 Multiple logistic regression model considering the outcome Fig. 3 LOS versus nutritional status—well nourished patients.
variable ‘complications’ Histogram representing the distribution of the number of well-
nourished patients and length of hospital stay (y-axis=number of
Risk factors OR1 Confidence interval patients; x-axis=days of length of stay; the curve is the distribution of
n
the mean time).
Malnutrition (moderately and severely) 1.60 1.09–2.35
Age Z60 years 1.71n 1.16–2.51
Presence of infection 1.71n 1.16–2.53 Table 5 Cox regression model to evaluate protective factors on LOS
1
OR=Odds ratio. Protective factors OR1 Confidence interval
n
Po0.05. Absence of complications 0.51n
0.42–0.62
Absence of cancer 0.80n 0.66–0.99
120 Well nourished 0.70n 0.59–0.83
1
OR=Odds ratio.
n
Po0.05.
100

The analysis of costs showed that malnourished

Number Of patients

80
patients represented a mean daily expense of US$
228.00/patient compared to the US$ 138.00/patient in
60 the well nourished. This represented an increased cost of
60.5% for malnutrition. When the costs of medications
40
and tests were added using respiratory infection patients
for comparison, the costs of the malnourished rose by
308.9% compared to the well-nourished patients.
20 Despite the high prevalence of malnutrition at
admission, nutritional therapy was prescribed to very
0 few patients. Enteral nutrition was used by 1.6% of all
0 40 80 120 160 200
the patients, parenteral nutrition by 0.8% and oral
Length of hospital stay in days
supplementation by 2.1%. Malnourished patients re-
Fig. 2 LOS vs nutritional status—well nourished patients. Histogram ceiving enteral nutrition were 4.9% of the total cases
representing the distribution of the number of malnourished patients
and length of hospital stay (y-axis=number of patients; x-axis=days and oral supplementation in 3.3%. Parenteral nutrition
of length of stay; the curve is the distribution of the mean time). was not prescribed to any malnourished patient.

contributing to shorter LOS. These were entered into Discussion

the Cox multivariate model. The statistically significant
results can be found in Table 5.
Hospital mortality was also significantly influenced by
the nutritional status of the patients, as shown in Table
6. Other factors were also found to influence mortality
rate, e.g. presence of cancer and infection, age above
60 years old and clinical treatment. Malnutrition was
considered an independent risk factor, which signifi-
cantly contributed to mortality (Table 7).
More than 20 years ago, hospital malnutrition was
described as highly prevalent although not well identi-
fied by medical teams (10). At the end of the 20th
century and beginning of a new millennium, hospital
malnutrition continues to be prevalent (12) and remains
a problem of medical attitude (12, 13). In our previous
study, we showed that 48.1% of 4000 patients were
malnourished (12). Despite this high occurrence, the
238 THE IMPACT OF MALNUTRITION

Table 6 Mortality vs nutritional status, sex, age, presence of infection, cancer and type of treatment
Mortality Survival RR (IC)
y
Malnourished 30 (12.4%) 212 (87.6%) 2.63 (1.55–5.27)n
Well nourished 22 (4.7%) 444 (95.3%)
Male 28 (7.9%) 328 (92.1%) 1.15 (0.68–1.95)nn
Female 24 (6.8%) 328 (93.2%)
Age Z60 years old 29 (12.6%) 201 (87.4%) 2.60 (1.54–4.40)n
Age o60 years old 23 (4.8%) 454 (95.2%)
With infection 25 (10.8%) 207 (89.2%) 1.90 (1.13–3.20)n
Without infection 27 (5.7%) 449 (94.3%)
With cancer 15 (11.6%) 114 (88.4%) 1.82 (1.03–3.21)n
Without cancer 37 (6.4%) 542 (93.6%)
Clinical treatment 42 (13.4%) 272 (86.6%) 0.19 (0.10–0.37)n
Surgical treatment 10 (2.5%) 272 (86.6%)
n
Po0.05. nnP=NS.
y
Moderately and severely malnourished.

Table 7 Risk factors associated with death, analyzed by a multiple re-
gression model
Risk factors OR1 Confidence interval
Malnutrition 2
1.87n
1.01–3.43
Presence of cancer 2.07n 1.03–4.15
Age Z60 years old 2.30n 1.26–4.21
Surgical treatment 0.16n 0.08–0.35
n
Po0.05.
1
OR=Odds ratio.
2
Moderately and severely malnutrition.
nutritional status of hospitalized patients was seldom
assessed by most medical professionals (12). Hospital
malnutrition is, in general, a consequence of several risk
factors, of which the disease per se is one of the most
important. The latter often leads patients to have
negative net nutrient intakes, i.e. nutrient intake less
than their requirements (12, 17). Other factors such as
age, social background and LOS also have a negative
impact on the nutritional status (12). All of these factors
can be prevented if diagnosed and treated early.
Although previous studies have shown the impact of
nutritional status on morbidity, mortality, LOS and
hospital costs, analysis by multiple logistic regression
has rarely been used (2, 4, 18–22). Similarly to these
studies, we were able to demonstrate that malnourished
patients had significantly higher incidence of complica-
tions (27.0% vs 16.8%), increased mortality (12.4% vs
4.7%), longer LOS (mean of 16.7 days vs 10.1 days)
and increased hospital costs. We used multiple logistic
regression as a tool to assess the role of malnutrition
alone in the patient’s outcome, knowing that other
variables such as presence of infection and cancer,
among others, could represent a source of bias. By using
this, we were able to show that malnutrition was indeed
an independent predictor of outcome.
In our study, it was interesting to note that surgical
patients had a decreased mortality. This might be due
to the fact that only patients with considered operable
disease underwent surgery and therefore were less likely
to die, in the hospital.
The negative impact caused by malnutrition on
patient’s outcome was long ago demonstrated. In 1936,
Studley (22) was one of the first physicians to show that
a 20% loss of usual body weight was correlated to a
significant increase in mortality rate of patients under-
going surgical treatment of duodenal ulcers. Buzby et al.
(1) demonstrated that malnourished surgical patients
had up to a 46% incidence of complications. Anker
et al. (19) showed that patients with congestive heart
failure, who were malnourished, had an increased
mortality rate. Length of hospital stay was significantly
longer in the study done by Von Meyenfeldt et al. (23),
while Robinson et al. (4) demonstrated a two-fold increase
in costs in those patients with one or more risk factors for
malnutrition at hospital admission.
Unfortunately, disease and nutrition interact whereby
the disease may cause secondary malnutrition or
malnutrition may adversely influence underlying disease
(17). This makes it difficult to conclude that malnutri-
tion alone leads to the patient’s worst outcome.
However, evidence coming from hunger strikers that
did not suffer from any disease, showed that when the
loss of 38% of body weight was reached, one-third of
them died (1). Keys et al. (21) measured the functional
changes in young men who underwent 24 weeks of semi-
starvation followed by a period of refeeding. Muscle
strength was decreased by 30% and depression score
rose by 30%. Refeeding reversed all of these changes.
Even short periods of fasting may lead to metabolic
dysfunctions. Ljungqvist et al. (24) showed that an
overnight fast in well-nourished individuals, scheduled
for surgery, led to peripheral insulin resistance, with
simultaneous negative nitrogen balance in the post-
operative period. Therefore, we can assume that
malnutrition is one of the risk factors associated to a
poorer outcome in conjunction with other risk factors.
In our study, the severe infectious complications, such
as sepsis (3.7% in the malnourished vs 1.1% in the well
nourished) and abdominal abscess (2.1% vs 0.4%),
might suggest that malnourished patients were those
with decreased inmunocompetence. Others (5, 6, 20)
have shown that infectious complications in malnour-
ished patients result from decreased inmunological
response (both humoral and cellular). On the other
hand, in our study, the non-infectious complications

having statistical significance were respiratory failure
(6.2% in the malnourished vs 1.3% in the well
nourished), cardiac failure (2.5% vs 0.6%) and cardiac
arrest (5.8% vs 1.5%). These results might be linked to
decreased muscular functional capacity together with
contractility dysfunction of the respiratory muscles and
loss of cardiac function, as shown by others (25, 26).
The results of some of the previously mentioned
studies could be questioned using evidence-based
medicine, which defends the use of randomized clinical
trials to support results and define concepts. However,
scientific discovery can be based on hypothetical-
deductive processes (27) since in the case of malnutri-
tion, it would be unethical to randomize one group of
patients to starvation and compare their outcome to the
group of fed patients.
In summary, as we enter the third millennium,
malnutrition might be one of the most important factor
that interferes in health and disease, and is the most
common disease in the hospital setting. Epidemiologists
define ‘common disease’ as having a prevalence above
10% (28). If nutritional status and disease determine
patient outcome, we have reason to be concerned.
Hence, the best decision is to treat the disease and
nourish the patient. Therefore, it is fundamental to
understand the significant role that nutritional therapy
plays in improving the outcome of those who cannot or
may not eat during their disease process, similar to what
hemodialysis represents to patients with renal failure
or ventilatory support to those with respiratory failure.
Therefore, nutritional assessment should be routinely
performed at admission in an attempt to reduce
nutrition-related complications.
Acknowledgements
We would like to acknowledge the Funda,cão de Amparo à Pesquisa do
Estado de São Paulo (FAPESP, grants # 98/01870-7 and # 98/0169-9)
and also Abbott International for the educational grants that
permitted the execution of this study. We would like to acknowledge
the regional coordinators and their research teams, including: Dr
Paulo Boente, Salvador; Dr Hélvio Chagas Ferro, Maceió; Dr Lúcio
Flávio Alencar, Recife; Dr S!ılvio Dantas, Natal; Dr Paulo Roberto
Leitão Vasconcelos, Fortaleza; Dr Jorge Alberto Langbeck Ohana,
Belém; Dr Alúisio Trindade Filho, Bras!ılia; Dr Álvaro Armando C. de
Morais, Vitória; Dr Ricardo Rosenfeld, Rio de Janeiro; Dr Edson
Lameu, Rio de Janeiro; Dr Eduardo E. M. Rocha, Rio de Janeiro;
Nut. Luciana Z. Coppini, São Paulo; Dr António Carlos Campos,
Curitiba; Nut. Bernadette Weber, Porto Alegre; Dr Maria Cristina
Silva, Pelotas; Dr Mauro Kleber Sousa e Silva, Belo Horizonte.
References
1. Allison S P. Malnutrion, disease, and outcome. Nutrition 2000; 16:
590–591
2. Chima C S, Barco K, Dewitt J L A, Maeda M, Teran J C, Mullen
K D. Relationship of nutritional status to length of stay, hospital
Submission date: 4 February 2002 Accepted: 3 December 2002
CLINICAL NUTRITION 239
costs, and discharge status of patients hospitalized in the medicine
service. Aliment Pharmacol Ther 1997; 11: 975–978
3. Reinhardt G F, Jyscofski J W, Wilkiens D G, Dobrin P B,
Mangan J E, Stannard R T. Incidence and mortality of
hypoalbuminemic patients in hospitalized veterans. J Parenter
Enteral Nutr 1980; 4: 357–359
4. Robinson G, Goldstein M, Levin G M. Impact of nutritional status
on DRG length of stay. J Parenter Enteral Nutr 1987; 11: 49–51
5. Chandra R K, Kumary S. Effects of nutrition on the immune
system. Nutrition 1994; 207–210
6. Van Der Hulst R R, Von Meyenfeldt M F, Van Kreel B K. Gut
permeability, intestinal morphology, and nutritional depletion.
Nutrition 1998; 14: 1–6
7. Efthimiou J, Fleming J, Gomes C, Spiro S G. Effect of
supplementary oral nutrition in poorly nourished patients with
chronic obstructive pulmonary disease. Am Rev Resp Dis 1988;
137: 1075–1082
8. Hill G L, Haydock D A. Impaired wound healing in surgical
patients with varying degrees of malnutrition. J Parenter Enteral
Nutr 1989; 10: 550–554
9. Weinsier R L, Hunker E M, Krumdieck D L, Butterworth D E.
Hospital malnutrition: a prospective evaluation of general medical
patients during the course of hospitalization. Am J Clin Nutr 1979;
32: 418–426
10. Butterworth C E. The skeleton in the hospital closet. Nutrition
Today 1974; 9: 4–8
11. Bistrian B R, Blackburn G L, Vitale J, Cochran D, Naylor J.
Prevalence of malnutrition in general medical patients. JAMA
1976; 235: 1567–1570
12. Waitzberg D L, Caiaffa W T, Correia M I T D. Hospital
malnutrition: the Brazilian national survey (Ibranutri): a study of
4000 patients. Nutrition 2001; 17: 575–580
13. McWhirter J P, Pennington C R. Incidence and recognition of
malnutrition in hospital. BMJ 1994; 308: 945–948
14. Pagano M, Gauvreau K. Principles of Biostatistics. Belmont,
Duxbury Press, 1993, 524p.
15. Detsky A S, McLaughlin J R, Baker J P et al. What is subjective
global assessment of nutritional status? J Parenter Enteral Nutr
1987; 11: 8–13
16. Buzby G P, Mullen J L, Matthews, D C. Prognostic nutritional
index in gastrointestinal surgery. Am J Surg 1980; 139: 160–167
17. Jeejeebhoy K N. Nutritional assessment. Nutrition 2000; 16: 585–589
18. Agrady E, Messina V, Campanell G et al. Hospital malnutrition:
incidence and prospective evaluation of general medical patients
during hospitalization. Acta Vitaminol Enzymol 1984; 6: 235–237
19. Anker S D, Ponikowsky P, Varney S et al. Wasting as independent
risk factor for mortality in chronic heart failure. Lancet 1997; 349:
1050–1053
20. Ek A C, Larsson J, Von Cchenck H, Throslun S, Unosson M,
Bjurrulf P. The correlation between eanergy, malnutrition and clinical
outcome in an elderly hospital population. Clin Nutr 1990; 9: 185–189
21. Keys A, Brozek J, Henschel A. The Biology of Human Starvation.
Minneapolis: University of Minnesota Press, 1950.
22. Studley H O. Percentage of weight loss. A basic indicator or
surgical risk in patients with chronic peptic ulcer. JAMA 1936;
106: 458–460
23. Von Meyenfeldt M F, Jeijerink W J H J, Rouflart J M J,
Builmaassen M T H J, Soeters P B. Perioperative nutritional
support: a randomized clinical trial. Clin Nutr 1992; 11: 180–186
24. Ljungqvist O, Nygren J, Thorell A. Insulin resistance in elective
surgery. Surgery 2000; 128: 757–760
25. Popper K R. The logic of scientific discovery. New York: Haper &
Row, 1968.
26. Zeiderman M R, McHahon M J. The role of objective measure-
ment of skeletal muscle function in the pre-operative patient. Clin
Nutr 1989; 8: 161–166
27. Heymsfield S B, Bethel R A, Asley J D, Gibbs D M, Felner J M,
Nutter D O. Cardiac abnormalities in cachectic patients before
and during repletion. AM Heart J 1978; 95: 584–594
28. Kahn H A, Sempos C T. Statistical methods in epidemiology. 2nd
ed. New York: Oxford University Press, 1989