Malnutrition is common in acutely ill patients, occurring in 30–50% of hospitalised patients [1, 2] a large proportion of who are already undernourished on admission or develop it while in hospital [3]. It is a serious public health problem and can adversely affect virtually every organ and/or system in the human body [4–7]. Malnutrition is associated with decreased muscle fat, decreased immune function, impaired quality of life, impaired wound healing, increase length of hospital stay, mortality and costs of health care [8–11]. It can thus be a health outcome as well as a risk factor for disease and exacerbated malnutrition [12]. The nutritional status of patients has been described in various terms. Some authors define it in terms of both undernutrition and overweight [7, 12, 13] while others use it to mean undernutrition alone [4, 14]. Undernutrition defined as a nutritional disorder status resulting from reduced nutrient intake or impaired metabolism [15].

Nutritional status is a multidimensional phenomenon that requires several methods of assessment. These include nutrition related health indicators which are anthropometric and biochemical data (BMI, serum albumin, prealbumin, haemoglobin, C reactive protein, calcium, magnesium, phosphorus), nutritional intake and energy expenditure [16, 17]. The use of a single assessment method to determine malnutrition in hospitalised patients has proved to be generally unsatisfactory. Attention has thus turned to combination of diverse measurements to increase sensitivity and specificity [18, 19]. Although these indicators can be epidemiologically useful, there is no gold standard for assessment of nutritional status [18, 20].

Despite the high prevalence and consequences of malnutrition, medical awareness of the patients nutritional status is still lacking [6, 21, 22] where as there is evidence that identifying and treating malnutrition in hospitalized patients is essential to improving outcomes [23]. The disease burden of malnutrition has been extensively documented in developed countries. There is only limited literature on nutritional screening and assessment in adults admitted to medical wards in sub-Saharan Africa. In addition, most health professionals seldom assess the nutritional status of hospitalised patients [22]. It was with this in mind that we aimed at screening for malnutrition using anthropometric and biochemical indices in patients admitted to the medical wards of the Douala General Hospital (DGH). We hope that by this we will improve our understanding and raise awareness of the problem of malnutrition, to enable care givers and policy makers to better meet the needs and improve outcomes of hospitalised patients [23, 24].

Malnutrition in hospitalized patients is common and has been extensively reported in the literature but still remains undetected and untreated in our setting. This is due to poor awareness and insufficient knowledge of the attending staff and policy makers [28, 29]. Recognition of malnutrition in hospitalised patients should remain a significant component of in-patient care as it has vast clinical and economic consequences [28]. We aimed at screening patients admitted to the Internal medicine wards in a tertiary medical centre in Cameroon using anthropometric and laboratory parameters in order to increase awareness of the burden of hospital malnutrition in similar settings. The outcome of this screening would then help define causes of action and intervention. From this study, malnutrition as defined by BMI and MUAC was present in 11.5 and 8.4% of patients respectively. BMI and MUAC correlated positively and significantly. 42.4% of patients reported weight loss greater than 10% in the 6 months before hospitalization. Blood albumin and hemoglobin were significantly lower in malnourished patients. Malnourished patients had a significantly longer LOS when compared to those with no malnutrition.

The proportion of our study population with malnutrition, defined by a BMI less than 18.5 kg/m² was 11.5%. This was similar to findings in other settings, Poland [30] and Scotland [31] using the same method of screening. However another Scottish study [21] found the prevalence of malnutrition in hospitalised patients to be four times higher than ours. While their study setting was different, patients were diverse, chosen from five different departments in the hospital. In two studies from western Nigeria, the prevalence of malnutrition was 7.8% [4] and 15% [32] respectively. In both, the participants were elderly and above 60 years of age. In the former study which was hospital based, 80% were outpatients and the screening tool was different whereas in the latter using BMI as us, the study was purely community based.

When we used MUAC to define malnutrition, we found a proportion of 8.4% of our subjects. Linear regression analysis showed strong evidence of a positive relation between BMI and MUAC (r = 0.78, P < 0.0001). These two indices agreed moderately and correctly classifying 92% of all cases. A similar correlation of these two indices has been found by other authors [33, 34]. In view of this, use of both BMI and MUAC can be acceptable for screening in our setting as they are practical, simple and cheap. MUAC could even be more useful in very ill patients who cannot be weighed or have their heights measured as was the case in this study. For further analysis, we classified malnutrition as having at least one of the two criteria. The reason for this was because BMI and MUAC showed a strong evidence in favor of an agreement, meaning that they could be interchangeable in classifying malnutrition. If we used the two methods in combination, malnutrition was present in 19.34% of the study population. A finding that 42% of participants presented with more than a 10% unintentional weight loss on admission was similar to an earlier study in similar settings [35]. Many authors found that when nutritional screening was assessed by anthropometric indices, large discrepancies may exist as in this study, thereby advocating for use of combination of indices as in validated screening tools [28]. These screening tools will definitely add one more task that may encumber already busy nursing or medical staff but will eventually have to be introduced for more complete assessments once extra training and nutritional support teams have been put in place.

Of all laboratory parameters used, serum albumin and hemoglobin were found to be significantly lower in patients diagnosed as malnourished. Serum albumin is a widely used variable for diagnosis and follow-up of patients with malnutrition [36]. However, though a poor indicator of body protein mass, it has been found to be correlated with non-anthropometric parameters and is a sensitive indicator of morbidity, mortality and length of hospital stay [17, 36, 37]. Serum albumin levels should not be relied on to define malnutrition but should remain a part of the nutritional status assessment in addition with the other indices [36]. Similarly, hemoglobin should be assessed to confirm for anemia which could be due to anemia of chronic disease or the nutritional anaemias (Iron deficiency, vitamin B₁₂, Folate deficiencies [38].

Several studies have repeatedly demonstrated the close relationship between disease and malnutrition [39]. If nutritional status and disease determine patient outcome, we have reason to be concerned. So as we treat the disease, we should nourish the patient [6]. In our practice, we generally screen for other features of the patients primary disease such as blood pressure, fever, dehydration and treat them whereas nutritional problems causing significant risk are unfortunately not identified [3]. We found that malnutrition was more common in patients with malignancies and chronic infections (HIV/AIDS and TB). In high income countries malignant disease has been shown to be a major contributor of malnutrition in hospitalized patients [11, 39] which was similar as in this study. However, a study in similar settings as ours, found a high frequency of malnutrition in routine clinical practice in HIV and TB patients in an area where these infections were highly prevalent [35].

Length of hospital stay is influenced by several factors which may include age, diagnosis, severity of disease, treatment and nutritional status [40]. Like many other authors, we also found that patients defined as malnourished had a significantly longer LOS compared to those who were not [39, 40]. Increased LOS is associated with higher costs for health care [6, 39, 41]. Our findings were consistent with most of the assessment methods and tools used to define malnutrition [30, 40].

Our study had some limitations. It was carried out in a single and tertiary centre which may present some referral bias with patient characteristic being different from other settings. Severely ill patients, who may have been the most malnourished, were not included as their weight and height could not be measured. Use of percentage weight loss as an index of malnutrition may have been inaccurate because previous usual weights were from patient recall. Lastly, concluding on the frequency of malnutrition in each disease entity may not be accurate due to the small numbers of patients in each group.

This study has shown that malnutrition is common in patients admitted to medical wards in the DGH. Serum albumin and hemoglobin levels were significantly lower in malnourished patients who also had a longer LOS than well-nourished patients. Nutritional screening and assessment should therefore be integrated in the care package of all patients admitted to hospital.