Risk factors associated with malnutrition in hospitalized patients

1 Instituto Federal de Educação, Ciência e Tecnologia do Amazonas, Setor de Saúde. Manaus, AM, Brasil. 2 Universidade Federal do Amazonas, Faculdade de Medicina, Departamento de Saúde Coletiva. R. Afonso Pena, 1053, Praça 14, 69025-050, Manaus, AM, Brasil. Correspondência para/Correspondence to: MC OLIVEIRA. E-mail: . Article based on the marter’s thesis of RFM FRAGAS, intitled “Prevalência da desnutrição em hospitais públicos de grande porte em Manaus AM”. Fundação Oswaldo Cruz; 2015. ORIGINAL | ORIGINAL


I N T R O D U C T I O N
The primary causes of hospital malnutrition are the underlying disease and treatment complications.However, some isolated factors or their interaction can increase the risk of hospital malnutrition and consequent undesirable or fatal outcomes 1 .
In addition to the epidemiological or prepathogenic aspects stemming from exposures before hospital admission, risk factors increase during the disease period and are very common in the hospital environment because they are inherent to the disease, symptoms, and treatment 2 .Among these risk factors, the hypercatabolic state provoked by conditions such as cancer or infectious diseases worsens nutritional status, as do long hospital stays, age higher than 60 years 3,4 , recent and involuntary weight loss 5 , changes in dietary consistency 6 , and gastrointestinal symptoms, such as nausea, dysphagia, abdominal pain, and anorexia, which somehow affect adequate food intake and nutrition.Thus, it is critical to identify the risk factors associated with malnutrition on hospital admission to try to avoid or minimize the deleterious effects on patients' recovery 7,8 , given that hospital malnutrition is associated with longer hospital stays and higher incidence of infections and postoperative complications 9 , further burdening the public health system.
In this context, the objective of this study was to determine the variables associated with a higher risk of hospital malnutrition in patients staying at public hospitals of Manaus, Amazonas, Brazil.

M E T H O D S
This is a secondary analysis of data from a quantitative epidemiological and hospital malnutrition study of three large general hospitals from the public health network of the city of Manaus (AM) from March to May 2014.
The eligibility criteria were: age equal to or greater than 18 years, either gender, and staying at the clinical, surgical, or intensive care wards of the hospitals.Individuals with plaster casts, submitted to limb amputations, or in clinical conditions that prevented anthropometric assessment were included in the study but only submitted to Subjective Global Assessment (SGA).
Pregnant or postnatal women and individuals with altered consciousness, unable to understand or communicate to answer the SGA questions and with no family member present to answer the questions were excluded.
The study included only the Sistema Único de Saúde general hospitals and emergency rooms with more than 150 beds that are directly managed by the Ministry of Health, Municipal Health Department, or State Health Department.
The researchers were trained for data collection, which included measuring height, weight, mid-upper arm circumference, physical examination, and filling out the study instrument.The SGA was performed by a single observer in order to minimize a possible diagnostic bias.
Body weight was measured by the portable bioimpedance scale Plenna Linea II ® (Plenna Especialidades Ltda, São Paulo, Brazil) with maximum capacity of 150 kg and height by the portable stadiometer WCS ® (Curitiba, Paraná, Brazil) with a total length of 220 cm and accuracy of one centimeter 10,11 .The Mid-Upper Dominant Arm Circumference (MUAC) was measured at the midpoint between the acromion and olecranon by an inelastic tape measure with the arm hanging by the side.The patients were assessed barefoot and with light clothing 10 .
The Body Mass Index (BMI) of all patients assessed by anthropometry was calculated 10,12 , and MUAC adequacy and classification relied on the cut-offs provided by Blackburn & Thornton 13 .Nutritional status diagnosis was based on the SGA model systematized by Detsky et al. 14 and adapted by Waitzberg & Ferrini 15 , classifying the individuals as well nourished, moderately malnourished, or severely malnourished.
Considering a 95% Confidence Interval (95%CI), the sample size was calculated by the program Epi info 7.0 (Centers of Disease Control and Prevention, Atlanta, Georgia, United States), based on the number of beds in each hospital plus 10% to compensate for eventual losses.Additionally, the estimate considered the prevalence of hospital malnutrition reported in the literature, which varies from 30 to 60% 16,17 , using a rough prevalence of 50%, reported by the Inquérito Brasileiro de Avaliação Nutricional Hospitalar, conducted in 1996 4 .
The sample size of the hospitals was initially obtained by mapping bed distribution and numbering them sequentially from one to the total number of beds of each hospital.Next, to determine the interval between the study patients, the ratio of the sampled population and the total number of beds was calculated, resulting in a systematic sample.
The literature was reviewed to select the variables that could be associated with nutritional status.The study variables were: age >60 years; long hospital stay (>15 days); cancer; infection; gastrointestinal diagnosis; neurological diagnosis; gastrointestinal symptoms; weight loss in the last 6 months; current weight loss; and persistent dietary change.
Gastrointestinal symptoms were defined as the presence of one or more of the following self-reported complaints for more than 15 days: dysphagia and/or odynophagia, anorexia, abdominal pain or distention, nausea, vomiting, and diarrhea.Regarding the diet, changes in consistency (fluid/soft) or quantity lasting more than 30 days were defined as persistent dietary changes.These changes were taken into account whether they occurred at the hospital, as part of a diet therapy, or before hospitalization, when patients changed their diets because of inappetence or eating difficulties in order to adapt to their new health condition.
The Odds Ratio (OR) was calculated to quantify the strength of association between nutritional status and the study variables, using a contingency table or 2x2.http://dx.doi.org/10.1590/1678-98652016000300003 Of the ten study variables, nine were included in the multiple logistic regression model because they were associated with malnutrition according to the Chi-square test, using as cut-off a p-value ≤0.20 or a 20% probability of error of the variable in question not being associated with malnutrition 5 .The strength of association between the variables in the final model was expressed as OR, with a 95%CI.The analyses were performed by the program Epi Info 7.0 18 .
The study was approved by the Research Ethics Committee of the Universidade Federal do Amazonas under Protocol nº 517,835, on January 29, 2014, and all participants and their buddies signed an Informed Consent Form before enrollment.

R E S U L T S
Of the 400 study patients from the three hospitals, one refused to participate in the study and two missed the examinations.Therefore, 397 patients were included.The mean age was 53±20 (18-95) years, and most were males (59.7%).
Among the risk factors associated with hospital malnutrition, most of the study covariates were highly prevalent in the malnourished patients, and some degree of association with this outcome was demonstrated by the OR, 95%CI, and p-value (Table 1).
The risk factors for hospital malnutrition with the highest associations were persistent dietary change and gastrointestinal symptoms, which were reported by most patients diagnosed with malnutrition.
Regarding the subjective global assessment, patients who reported involuntary weight loss in the last 6 months or those who reported current and continuous weight loss during the assessment were four times more likely to be malnourished than those who denied having had these occurrences.
The results also showed that older adults and individuals with more than 15 days of hospital stay were three times more likely to become malnourished than adults or those with fewer than 15 days of hospital stay.
Regarding the underlying disease, neurological disease doubled the chance of malnutrition compared with other medical diagnoses.Cancer patients were four times more likely to be malnourished, and infection was present in more than half of the malnourished individuals.In multivariate logistic regression (Table 2), the variables that remained associated were: persistent dietary change, gastrointestinal symptoms, cancer, current weight loss, weight loss in the last 6 months, and age >60 years.
Persistent dietary change was the most important variable for the risk of malnutrition.

D I S C U S S I O N
The study frequency of malnutrition was very similar to that found by other Brazilian and foreign studies 16,19,20 , corroborating the statement that along the last four decades, the prevalence of malnutrition has remained high 21 , increasing the length of hospital stay and risk of death 22 .Thus, it is very important to identify and monitor factors that possibly contribute to malnutrition during hospitalization in order to minimize complications.Regarding long hospital stays, studies 3,4 have shown that patients with more than 15 days of hospital stay were three times more likely to be malnourished (OR=3.09;95%CI=2.55-3.74;p<0.05;OR=3.00; 95%CI=2.61-3.45;p<0.05), corroborating the results from the Chi-square test, where length of hospital stay higher than 15 days tripled the odds of patients becoming malnourished compared with shorter hospital stays.However, in multivariate analysis this variable lost strength of association because other factors proved to be more important with respect to malnutrition.
Regarding these factors, multiple regression analysis identified predictive variables, that is, variables associated with malnutrition in the study sample.
The subjective global assessment proposed by Detsky et al. 14 , a method widely used in clinical practice, consists of an anamnesis about recent body weight changes, changes in dietary patterns, and gastrointestinal changes.Given the patients' observations regarding food intake, the main study risk factor was persistent dietary change, that is, a change maintained for more than 30 days.Ferreira et al. 23 found that decreased food intake is frequently observed in hospitalized patients, which can be related to dissatisfaction with hospital food and disease symptoms.According to Waitzberg et al. 24 , clinical symptoms like diarrhea, nausea, vomiting, and dysphagia generally lead to loss of appetite, food refusal, or food intake difficulties, which can contribute to nutritional status non-recovery or deterioration. .
Aquino & Philippi 8 found that 62% of malnourished individuals reported changes in quantity, type, and consistency of the foods consumed secondary to the very frequent gastrointestinal changes, such as nausea, vomiting, and gastric changes.Consequently, the researchers found that recent and involuntary weight loss was the most important variable associated with risk of malnutrition (OR=58.03;95%CI=18.46-182.41;p<0.001).Hence, determining the influence of hospital food and the factors and symptoms that affect food intake and contribute to malnutrition in hospitalized patients may improve nutritional intervention actions and have a positive impact on length of hospital stay 1 .
In multivariate analysis gastrointestinal or neurological diseases did not remain associated with malnutrition in the study sample.Nonetheless, cancer patients were more likely to become malnourished than those with other diagnoses, which corroborates Waitzberg et al. 4 , who considered cancer a risk factor for malnutrition because they found that the frequency of malnutrition was three times higher in hospitalized patients with cancer.
Cancer patients' higher vulnerability is due to higher metabolic expenditure, protein deprivation caused by the tumor 26 , and the side effects of the treatment of choice.According to Ferreira et al. 23 , symptoms like anorexia, xerostomia, mucositis, and nausea are closely related to low acceptance of hospital food and consequently, with a higher risk of malnutrition.
Considering age as a risk factor for malnutrition, the results of this study corroborate results described elsewhere 3,4 , showing that changes associated with ageing promote changes in the nutritional requirements 27 .Older adults are a group at risk of malnutrition because of a number of morphological, motor, and sensory changes throughout the entire gastrointestinal tract, which has an important role in the regulation of appetite, satiety, and dietary pattern 28 .
The frequency data exposed in Table 1 also show that more than half of the malnourished individuals were older adults.Hence, given the high prevalence of malnutrition in older adults, Fidelix et al. 29 reported the need of implementing strategies to identify, prevent, and treat this condition both in the hospital setting and after discharge.
One of the limitations of the method used is the reliance on the individual's memory and disposition to provide the necessary information for the association analyses.Additionally, it is important to note that the process of malnutrition is continuous and other variables may not have been considered because of the cross-sectional nature of the study, which prevents the inference of a cause-effect relationship.

C O N C L U S I O N
The analyses suggest two important variables that deserve attention during hospitalization, persistent dietary changes and gastrointestinal symptoms.Generally, hospitalized patients need routine dietary follow-up between hospital routines and the identification of symptoms that may affect food acceptance, which can be easily determined by subjective assessment.Moreover, hospital admission should include body weight measurement and inquiries about recent weight loss.
Thorough knowledge about the current prevalence of malnutrition and the factors that contribute to nutritional status deterioration in hospitalized patients is critical in order to improve the quality and efficiency of nutritional care, given that it is a vital component for the patient's recovery.

C O N T R I B U T O R S
RFM FRAGAS conceived and designed the study, collected, analyzed, and interpreted the data, discussed the results, and wrote the manuscript under the supervision of MC OLIVEIRA.

Table 1 .
Chi-square test of the variables possibly associated with malnutrition listed in decreasing Odds Ratio (OR).

Table 2 .
Multiple logistic regression analysis of the variables associated with malnutrition at a significance level of 5% (p≤0.05) after the Chi-square test.Manaus (AM), Brazil, 2014.