The relationship between malnutrition risk and clinical outcomes in a cohort of frail older hospital patients

Highlights

• The ‘MUST’, may lack diagnostic accuracy in frail older hospital patients.

• The ‘MNA-SF’ may demonstrate greater accuracy and predict mortality risk.

• A potential BMI mortality paradox is highlighted in this patient group.

• Fat mass and fat mass index measurements may predict mortality risk.

• A combination of assessment methods may be clinically useful.

Summary

Background & aims

Malnutrition has an adverse effect on clinical outcomes and frail older people may be at greater risk of malnutrition. The purpose and aims of this study was to investigate the relationship between markers of malnutrition risk and clinical outcomes in a cohort of frail older hospital patients.

Methods

78 frail older hospital patients had the following measurements recorded; length of stay (LOS), time to medical fitness for discharge (TMFFD), body mass index (BMI), malnutrition universal screening tool (MUST) and mini-nutritional assessment short-form (MNA-SF) scores, blood urea, C-reactive protein (CRP), albumin, CRP-albumin ratio; and bioelectrical impedance assessment (BIA) measurements (n = 66). Patients were grouped by mortality status 12 months post hospital admission. Grouping by albumin classification was performed (n = 66) whereby, <30 g/l indicated severe malnutrition, 30–34.9, moderate and >35, low. Receiver-operating characteristic (ROC) curve analysis was performed on variables as potential predictors of mortality.

Results

After 12 months, 31% (n = 24) of patients died. LOS was significantly greater in this group (25.0 ± 22.9 vs 15.4 ± 12.7d, P < 0.05). BMI (23.8 ± 4.9 vs 26.4 ± 5.5 kg/m²); fat mass (FM) (17.2 ± 9.9 vs 25.5 ± 10.5 kg), fat mass index (FMI) (9.3 ± 4.1 vs 17.9 ± 2.4 kg/m²); and MNA-SF score (6.6 ± 2.4 vs 8.6 ± 2.7) were significantly lower (P < 0.05), and urea significantly higher (11.4 ± 8.7 vs 8.8 ± 4.4 mmol/l, P = 0.05). Albumin was typically low across the entire group (30.5 ± 5.9 g/l) and a potential relationship was identified between albumin and MNA-SF score. MNA-SF, FM, and FMI were significant predictors of mortality outcome by ROC curve analysis, whereas MUST was a poor predictor.

Conclusion

This study highlights a potential relationship between indicators of malnutrition risk and clinical outcomes in frail older hospital patients which should be studied in larger cohorts with an aim to improve patient care.

Introduction

Frail older people may be admitted to hospital wards suffering from a range of acute and chronic disease/s, with signs and symptoms of physical and/or cognitive frailty and be on multiple medications. Identifying possible nutritional risk/malnutrition is important and may affect trajectory of health, morbidity, and mortality [1], [2], [3], [4]. Different screening methods exist including the ‘malnutrition universal screening tool’ (MUST) [1], [5], the ‘mini-nutritional assessment’ (MNA) [1], [6], [7], [8] and the ‘geriatric nutritional risk index’, (GNRI) [9]. In the United Kingdom (UK), the MUST is the standard routine method of screening in all hospital wards and care homes, although in reality there is no universal gold standard tool [4]. We showed recently in a cohort of frail older hospital patients that there is a significant discordance between MUST and ‘MNA-short form’ (MNA-SF) malnutrition screening categorisation [10]. The MUST predominantly categorized patients as ‘low risk’ (77%) and MNA-SF predominantly as ‘at risk’ (46%) and ‘malnourished’ (45%). Reliability assessment found poor reliability between the screening tools and bioelectrical impedance assessment (BIA) assessment was in general agreement with MNA-SF scoring patterns, especially in male patients. A potential body mass index (BMI) paradox was also highlighted whereby some patients who were ‘at risk’ or ‘malnourished’ by MNA-SF scores had normal BMI and depleted/borderline BIA measurements of fat free mass (FFM)/fat mass (FM) and specifically indices (FFMI and FMI, in kg/m²). Potential reasons for the observed MUST-MNA-SF discordance include: the MUST uses World Health Organization (WHO) BMI grading criteria, and there maybe difficulty in obtaining accurate weight loss information in this patient group. Further, the MNA-SF has additional screening questions on ‘mobility’ and ‘neuropsychological problems’ which would create a tendency to score worse in a frail older patient group.

An important area to address which overlaps malnutrition is ‘cachexia’/‘cachexia-risk’, as acute and chronic illness has a typical effect upon food intake (anorexia) and metabolism (e.g. hypermetabolism and raised protein breakdown), principally through actions of circulating proinflammatory cytokines [11], [12]. Other measurable domains of nutritional status which are sensitive to malnutrition and inflammation include important blood markers such as albumin, which is utilised in the GNRI [9], and is a well known prognostic marker [13], [14], [15], [16]. C-reactive protein (CRP) is another routine blood marker indicating inflammatory status and has known prognostic potential [17], [18]. Recently, the CRP/albumin ratio has been used to better predict mortality risk in septic patients [19].

A better understanding of the relationship between malnutrition risk screening, body composition assessment and blood markers in heterogeneous groups of frail older hospital patients on clinical outcomes may improve coordinated hospital nutritional care in the UK.

This study was undertaken in a heterogeneous group of frail older adults admitted to wards specialising in elder care in the UK. We examined outcome of hospital admission, length of stay (LOS), time to medical fitness to discharge (TMFFD) and mortality at 12 months post admission and related them to inpatient measurements of MUST, MNA-SF and BIA. Further, examination was made of routine blood markers, urea, albumin, CRP, and the CRP/albumin ratio to investigate their importance in relation to malnutrition risk and outcomes.