Scored patient-generated Subjective Global Assessment: Length of hospital stay and mortality in cancer patients

1 Universidade Federal do Maranhão, Centro de Ciências Biológicas e da Saúde, Programa de Pós-Graduação em Saúde do Adulto e da Criança. Av. dos Portugueses, 1966, Prédio das Pós-Graduações, Bacanga, 65085-580, São Luís, MA, Brasil. Correspondência para/Correspondence to: AF SANTOS. E-mail: . 2 Universidade Ceuma, Coordenação do Curso de Nutrição. São Luís, MA, Brasil. 3 Faculdade Santa Terezinha, Coordenação do Curso de Nutrição. São Luís, MA, Brasil. 4 Universidade Federal do Maranhão, Centro de Ciências Biológicas e da Saúde, Departamento do Curso de Nutrição. São Luís, MA, Brasil.


I N T R O D U C T I O N
Malnutrition is a fairly common finding in cancer patients.In Brazil, a study carried out in 45 centers of excellence for cancer diagnosis and treatment estimated a prevalence of malnutrition of 45.1% among the almost five thousand patients evaluated [1].
Factors commonly associated with malnutrition in cancer patients include anorexia, chewing and swallowing difficulties, side effects of cancer treatment, and excessive fasting for tests or procedures [2].Other factors include low socioeconomic status and previously inadequate eating habits [3].
Thus, malnutrition and cancer tend to lead to reduced food intake, metabolic changes, and increased nutrient demand caused by tumor growth [2].
The impact of this process is expected to encumber the health services and is largely https://doi.org/10.1590/1678-98652017000500001associated with negative outcomes, such as: increase in morbidity and mortality, infections and complications, length of hospital stay (six days or over), and hospital costs and decrease in healing capacity, immune response, and tolerance to anticancer treatment [4,5].
The scored Patient-Generated Subjective Global Assessment (PG-SGA) has been considered as the standard method of nutritional assessment for patients with cancer in several robust studies comparing it with different methods of nutritional assessment [3,[5][6][7][8].It is a practical method that allows quick identification and includes low cost, objective, and subjective assessment instruments [9].Thus, the use of the PG-SGA combined with a protocol that includes cancer treatment can screen patients who may benefit from a specific intervention [7,9].This is a subjective method composed of questions about changes in weight and dietary intake, gastrointestinal symptoms, and functional capacity, which are answered by the patient.It also includes a form containing data on increased nutritional needs due to the disease, metabolic demand, and physical examination.Each item is scored, and the sum of the scores obtained in each domain is classified according to the following SGA classification: SGA A (wellnourished), SGA B (moderately malnourished) and SGA C (severely malnourished) [9].
Therefore, the present study aimed to associate the scored PG-SGA with mortality and length of hospital stay in cancer patients.

M E T H O D S
This is an analytical cross-sectional study carried out between July and September 2014 using secondary data collection using data from 366 medical records of patients admitted to a hospital recognized as a cancer center of excellence between January 2010 and January 2014.
Patients diagnosed with malignant neoplasm, aged at least 20 years and with a minimum hospital stay of and 3 days, and who had been evaluated using the PG-SGA were included in this study.Patient gender was randomly chosen.Medical records of pregnant women or those that were illegible or with incomplete data were excluded.All eligible medical records were selected using probability sampling.
The calculation of sample size was based on a total of 1,193 PG-SGA forms registered at the Hospital's Nutrition and Dietetics Service between January 20110 and January 2014, with a statistical power of 95%, margin of error of 5%, and a prevalence of malnutrition of 45.1% among cancer patients [1]; 15% of sample loss were added to the calculation resulting in a minimum sample of 333 patients.

Patient-Generated Subjective Global
Assessment was the basis of information collection (anthropometric data, changes in dietary intake, presence of symptoms, functional capacity, disease data and their association with nutritional needs, and physical examination).This questionnaire incorporates a numerical score and a global rating, in which higher scores are indicative of greater nutritional risk.In the present study, the classification of the instrument was divided into scores as follows: score <17 referred to well-nourished patients; scores ≥17, referred to patients with some degree of malnutrition; score <22 referred to moderately malnourished patients; and score ≥22 referred to severely malnourished patients [9].Clinical data were collected from the medical records: age; gender; place of origin (city of São Luís, state of Maranhão, Brazil), Metropolitan Region, or inland cities in the state); disease diagnosis according to the International Classification of Diseases; cancer treatment (chemotherapy, radiotherapy, combined treatment, surgical treatment); length of hospital stay (≥10 days); length of administration of PG-SGA ≥2 days (yes or no) and length of administration of PG-SGA ≥5 days (yes or no); patient outcome (hospital discharge or death); weight loss in one and six month-periods; Body Mass Index (BMI); and total PG-SGA score.It is worth emphasizing that in terms of BMI, the patients were classified as: malnourished (BMI <18.5kg/m² for adults and <22.0kg/m² for elderly patients), well-nourished (BMI≥18.5-24.9kg/m²for adults and ≥22.0-27.0kg/m²for elderly patients) and overweight/ obese (BMI ≥25.0kg/m² for adults and ≥27.0kg/m² for elderly patients) [ The Chi-square test was used to evaluate the associations between the frequency (absolute and relative) of variables.Adjusted and unadjusted logistic regression models were used to analyze the independent and dependent variables, employing the Backward method.Data were shown in tables, and all statistical associations were considered as significant when alpha value was <0.05.
The study was approved by the Research Ethics Committee of the Hospital Universitário of Universidade Federal do Maranhão (Protocol nº 711.819/2014).

R E S U L T S
There were no statistically significant associations between the presence of malnutrition and gender, age, and place of origin.There was a high frequency of malnutrition (36.0%).The highest prevalence of malnutrition was found among women (40.2%) aged 59 years or older (40.6%) and who were from the metropolitan region of Maranhão state (45.8%), according to Table 1.
There was higher frequency of malnutrition (37.6%) among individuals with tumors in the digestive system.The presence of metastasis was statistically associated (p=0.019) with a higher frequency of malnutrition (52.4% of the patients).There was a statistically significant (p<0.001)higher frequency of malnutrition (55.0%) among the patients who had died ( Unadjusted logistic regression was carried between the variable length of hospital stay (≥10 days) and the variables of interest (gender, age, primary tumor location, treatment, length of administration of PG-SGA of two and five days, presence of some degree of malnutrition, presence of severe malnutrition, and type of discharge).There was statistical significance (p<0.05)only in the association of the combined treatment (radiotherapy and chemotherapy) with five-day administration of PG-SGA.
Adjustment was applied to factors of the logistic regression and it was found that length of administration of PG-SGA of up to 5 days and surgical treatment were significantly associated (p<0.05) with a reduction in the length of hospital stay (≥10 days).The administration of PG-SGA of up to 5 days led to 91% reduction in the cases of hospital stay ≥10 days.The combined treatment (radiotherapy and chemotherapy) was associated with an increase in the length of hospital stay (odds ratio=4.12,p=0.015).Similarly, the presence of severe malnutrition according to the PG-SGA (PG-SGA score greater than or equal to 22) indicated at least five times greater risk of longer hospital stay (≥10 days), p=0.047 (Table 3).
The unadjusted logistic regression for mortality showed that it was statistically significantly associated (p<0.05)only with tumors located primarily in the male reproductive system, presence of metastasis, clinical and combined treatments (radiotherapy + chemotherapy), and presence of malnutrition (PG-SGA ≥17).
After carrying out adjusted logistic regression, only the primary diagnosis of the disease in the male reproductive system, presence of metastasis, clinical treatment, longer hospital stay, and presence of malnutrition were  4).
The primary diagnosis of the disease in the male reproductive system was associated with a lower frequency of death (p=0.012).On the other hand, the presence of metastasis and malnutrition, the type of clinical treatment, and prolonged hospitalization were statistically associated (p<0.05) with an increase in mortality (odds ratio >2.0), Table 4.

D I S C U S S I O N
The prevalence of malnutrition found in this study (36.0%) is similar to that reported in the Brazilian Oncological Nutrition Inquiry, which showed a frequency of 45.1% of cancer patients in similar situation [1].
A multicenter research trial carried out in Latin American countries [10] reported much higher frequencies (65.6%) in cancer patients than those found in the present study.Its findings showed that among the factors associated with malnutrition according to the SGA (traditional form) are age >60 years, presence of infection, long hospital stay, and presence of cancer, which was the factor with the greatest impact on the patients evaluated.
Fragas & Oliveira [4], investigating factors associated with malnutrition during hospitalization, found that there was a 3-fold risk of malnutrition in prolonged hospitalizations (up to 15 days).Furthermore, persistent changes in diet, presence of digestive symptoms, recent weight loss and weight loss over the last six months, age, and mainly cancer (tenfold risk) also increase the risk of becoming malnourished.These data indicate that the factors associated with malnutrition can be first investigated when the patients are admitted to a hospital and can be used in an evaluation that contributes to an adequate nutritional support and recovery.
The different degrees of malnutrition in patients with cancer may indicate the different methods of assessment used, which should consider the reality of the hospital or health care center, usefulness, cost, speed of use, and especially the sensitivity of the nutritional screening tool.
Very different data from those obtained in the present study were reported by Bauer et al. [11], who found 76% prevalence of some degree of malnutrition, according to the PG-SGA.
In the present study, sociodemographic factors may have also contributed to the increase the prevalence of malnutrition since a significant number of patients were from inland cities the state of Maranhão.These patients may seek health care at a later stage, which may be associated with a significant worsening of their nutritional status due to the disease.
Lenders et al. [12] pointed out that the lack of medical awareness regarding patients' nutritional status is an important aggravating factor of malnutrition in the hospital environment.This issue was not the focus of the present study; however, questions about the formal medical education and training should be raised since they may not give due importance to the patients' adequate nutritional.Thus, they should be discussed and revised.
Correia et al. [13] conducted a systematic review of almost 70 Latin American studies (totaling more than 29 thousand studies evaluated) to analyze the effect of malnutrition during hospitalization.The authors pointed out that in addition to a high prevalence (between 40% and 60%) on admission to the hospital, there was an increase in the number of malnourished patients with an increased length of hospital stay.This was mainly associated with increased infectious and noninfectious complications, length of hospital stay, and associated hospital costs.
Adjusted logistic regression was used in the present study to identify the isolated impact of malnutrition, according to the PG-SGA, in two important hospitalized patient outcomes: mortality and prolonged hospitalization.
It should be noted that the presence of malnutrition according to the PG-SGA was associated with death (p<0.001), which may indicate a risk predictor in this population.The interpretation of this finding is important because it refers to the negative impact of malnutrition alone, evaluated using a practical, inexpensive, and rapid method.
The presence of malnutrition was also associated with long hospital stay, which may implicitly contribute to inadequate bed occupancy and exposure to nosocomial hospitalacquired infections and morbidities and, consequently, higher associated hospital costs.This may affect public health budgets in Brazil.
The high rate of malnutrition among patients with metastatic cancer (p=0.019) may reflect the effect of tumor staging on the nutritional status of cancer patients.Studies [14,15] have already shown that more advanced stages are important nutritional risk factors in the increase of catabolism in cancer patients.
Vigano et al. [7] found that cancer patients with higher PG-SGA scores (≥9 vs 0 to 1) had unfavorable biological markers of cancer cachexia, such as: high white blood cell counts, C-reactive protein, and lower hemoglobin levels, BMI, and fat mass.In addition, there was 12% greater length of hospital stay, a reduction in the doses of chemotherapy, and increased mortality.
Bauer et al. [11] found an association between the presence of malnutrition (PG-SGA scores ≥17) and an increase in the length of hospital stay, similar to the results obtained in the present study.However, unlike our findings, these authors did not find significant association with increased mortality.
It is worth mentioning that the use of subjective nutritional assessment instruments has been shown to be strongly associated with the length of hospital stay of cancer patients [14].
Tappenden et al. [17] emphasized the effects of an early nutritional intervention, which may contribute to a reduction in the complication rates, length of hospital stay, readmission rates, mortality, and cost of care.Accordingly, the authors argued that nutritional care should be based on six basic principles: (1) creation of an institutional culture with the collaboration of all health professionals involved in the nutritional care of patients; (2) redefinition of the role of clinical care professionals in nutritional care; (3) identification of all malnourished patients and those at nutritional risk; (4) implementation of periodic nutritional interventions; (5) sharing of nutrition care plans between all sectors involved; (6) development of a comprehensive nutrition care and continuing education plan.
Malnutrition and nutritional screening estimates can contribute to the planning of public health strategies for an early and comprehensive nutritional screening in cancer patients, who may become weak and vulnerable to malnutrition due to several factors reported in the literature, such as: tumor location [15], age [11], loss of fat and lean mass [18], and side effects to cancer treatment [19], among others.

C O N C L U S I O N
The Patient-Generated Subjective Global Assessment score is an important risk marker of prolonged hospitalization and high mortality rates.This tool proved to be useful and capable of circumventing significant biases in the nutritional evaluation of cancer patients.Nutritional impact measures should be taken to greatly reduce malnutrition rates.

A
C K N O W L E D G E M E N T S The authors are grateful for the financial support provided by the Fundação de Amparo à Pesquisa e Desenvolvimento Científico do Maranhão and to the Aldenora Bello Cancer Hospital.C O N T R I B U T O R S AF SANTOS and MBC CHEIN contributed to the project and conception of this study and to data tabulation and discussion of results.AA RABELO JUNIOR and FLB CAMPOS contributed to data collection and tabulation, discussion of results and conception of the study.RML SOUSA and HJF VELOSO contributed to the discussion of results and conception of the study.
2]. Patient-Generated Subjective Global Assessment scores were tabulated using Microsoft Office Excel ® spreadsheets, version 2013 (Microsoft Corporation, Redmond, Washington, United States).Data analysis was carried out with Stata ® version 12.0 (Stata Corp LP, College Station, Texas, United States).The Shapiro-Wilk test was used to test the normality of the variables.

Table 2 .
Association between the presence of malnutrition according to the scored patient-generated Subjective Global Assessment and characteristics of disease' diagnostic and hospitalization among cancer patients.São Luís (MA), Brazil, 2015.

Table 3 .
Factors associated with length of hospital stay greater than ten days in patients of a hospital recognized as a cancer center of excellence: adjusted logistic regression.São Luís (MA), Brazil, 2015.