The predictive value of weight evolution in screening for malnutrition in community-dwelling older persons (70+) in Antwerp

In a world where there is increasing attention for obesity and the associated health risks, it seems paradoxical to highlight malnutrition. However, malnutrition is a heavily underestimated and frequently occurring pathology [1]. In 2013, 7% of the over-70s living at home in Belgium were found to be malnourished [2]. In addition, 29% of this population had a risk of malnutrition [2, 3]. Due to the aging of the population and the increasing survival rate for a number of pathologies, the current prevalence is estimated to be even higher [1, 4]. Furthermore, malnutrition in a home situation will increase due to the decrease in the length of hospitalization and the shift of care to primary care [1].

The negative consequences of malnutrition cannot be underestimated. Studies showed more frequent disease episodes, a slower recovery, a higher risk of falling, a longer hospital stay, more complications, a reduced quality of life and a shorter life expectancy in patients with malnutrition [2, 4]. Depression and self-neglect are considered as consequences of malnutrition [3]. The economic impact of untreated malnutrition in Belgium was estimated at 400 million euros per year in 2005 [5]. Both the biopsychosocial consequences and the economic consequences can be prevented if (the risk of) malnutrition is timely detected and treated [6].

Screening for malnutrition is indispensable but health care providers only recognize half of malnourished patients [2, 7]. Various screening instruments are available to be used in primary care to detect (the risk of) malnutrition in an early stage. In a systematic review, the Mini Nutritional Assessment short form (MNA-SF) is considered the most suitable screening tool for community-dwelling older adults [8]. This test consists of 6 questions, can be performed in 5 min, and has a sensitivity of 98% and a specificity of 94% [9]. Nevertheless, this screening instrument is rarely used in primary care. The tool is insufficiently known by the general practitioner and is seen as time consuming during a routine consultation. Above, there are other practical barriers to pay sufficient attention to prevention [10, 11]. A multidisciplinary approach could make screening for malnutrition more efficient [1, 5].

Practical experience and research show that screening in primary care mainly takes place by monitoring weight and that validated screening instruments are hardly used [5, 10, 11]. However, care providers intuitively indicate that weight loss could be misleading to assess the nutritional state of patients [12]. Therefore, in this study we want to examine how effective and predictive weight evolution is in screening for (risk of) malnutrition in older people living at home, in comparison with a validated screening tool, namely the MNA-SF.

In the autumn of 2020, we contacted seven different groups of home nurses from the province of Antwerp, Belgium. Each group committed to this study and selected patients who met the inclusion criteria. In total, 143 patients gave their informed consent to participate, of which 89 were women and 54 men. The mean age was 83.7 years (SD 6.62) with a range of 70 to 100 years. (see Table 1).

Based on the MNA-SF score measured after six months, 53.1% (76/143) of participants had a normal nutritional status, 37.8% (54/143) was at risk of malnutrition and 4.9% (7/ 143) was malnourished. The MNA-SF could not be administered to six participants because of death, hospitalization or admission to a residential care center. Of the participants with (risk of) malnutrition, 44.3% (27/61) had a BMI < 23 kg/m², which is defined as underweight for older people [13]. From the participants with a normal score on the MNA-SF was 6.6% (5/76) underweight. The malnourished population had a mean BMI of 20.0 (SD 4.02), compared to 29.6 (SD 5.00) in the population with a normal score on the MNA-SF.

Of the total population, 32.9% (47/143) had a psychiatric multimorbidity. 70.2% (33/47) of people with psychiatric multimorbidity had (risk of) malnutrition. With regard to own nutritional assessment, 83.6% (51/61) of the participants with (risk of) malnutrition believed that they were well nourished and 78.7% (48/61) of this group indicated that they enjoyed eating moments, compared to respectively 92.1% (70/76) and 92.1% (70/76) of the participants with a normal score on the MNA-SF. (see Table 1).

In order to detect people with (risk of) malnutrition, this study showed a PPV of 78,6%, a NPV of 60,7%, a sensitivity of 19.3% and a specificity of 96.0% for a weight evolution of ≥ 5% weight loss over six months. To detect malnutrition, our results showed respectively 33,3%, 98,4%, 71.4% and 92.3%. (see Table 2; Fig. 1).

These results show that weight evolution in people over 70 living at home is a poor predictor in screening for (risk of) malnutrition. Compared to a validated screening tool, the MNA-SF, a weight loss of 5% or more appears to have a sensitivity of only 20% in detecting (risk of) malnutrition. However, a high specificity is observed (96.0%). A possible explanation for the low sensitivity is that the reduction in food intake is already present before weight loss occurs. Severe weight loss would occur too late to function as a nutritional marker in the screening for (risk of) malnutrition in clinical practice [15]. However, weight loss does seem to be a better predictor for detecting malnutrition, with a sensitivity and specificity of respectively 71.4% and 92.3% for a weight loss of 5% or more in this study group.

In this study, the prevalence of malnutrition is 5% and 38% of the participants are at risk of malnutrition. In 2013, the Nutri Action II study reported a prevalence of 7% and 28% respectively among 819 older persons living at home in Belgium [2]. The higher prevalence can partly be explained by the aging of the population accompanied by the accumulation of chronic conditions and the improved survival rates of certain conditions [1].

When developing the MNA-SF as a screening tool, a sensitivity and specificity of 79.5% and 77.4% were described for a BMI < 23 kg/m² as compared to the results on the full MNA [9]. BMI can help to estimate nutritional status but is less accurate as a screening tool than the MNA-SF, which has a sensitivity of 97.8% and a specificity of 94.3% [9].

Our results show that 70.2% (33/47) of people with psychiatric disease have (risk of) malnutrition. This observation supports the finding that the prevalence of (risk of) malnutrition appears to be higher in older persons with dementia or depression as compared to other co-morbidities [2].

A strength of this study is that the predetermined sample size was achieved. In addition, due to the multidisciplinary design and the size of the study population, many different healthcare providers are involved in this study. Another strength of the study is the use of the MNA-SF, a validated and well-developed screening tool for malnutrition for older adults [9]. In addition, patients who were diagnosed with a (risk of) malnutrition during the study were referred to their own GP for treatment and follow up.

A limitation of the study is that we included the home nurses via a convenience sample without randomization. In addition, mainly patients from rural areas were included in this study. Home nurses reported that patients in cities were more reluctant to give their informed consent. This may affect the representativeness of the study population. Furthermore, there sample of home nurses was very diverse, which might compromise the homogeneity. A selection bias is possible when nurses did not complete the study for all their patients who met the inclusion criteria or because certain patient groups refused to participate. The nurses might have deliberately selected patients based on their expected nutritional status, or presented the study only to the more cooperative patient, resulting in a lower prevalence of (risk of) malnutrition. The inclusion of people with dementia might also complicate the assessment of certain questions. Finally, the results referring to the malnutrition category in this study should be interpreted with caution, as the sample size was calculated on the basis of the prevalence of (risk of) malnutrition.

Weight evolution has a low sensitivity in screening for (risk of) malnutrition in people over 70 living at home as compared to the MNA-SF. In order to detect people with malnutrition, weight loss seems to have a considerable predictive value in this study.

To reliably define sensitivity and specificity of weight loss in the screening of malnutrition, a larger population is required with a sample size calculated on the basis of the prevalence of malnutrition.

Since the MNA-SF is considered to be the best screening tool for malnutrition in community-dwelling older adults, it is interesting to make this screening tool more widely known among primary care providers. This way, the focus will shift from only monitoring weight evolution to a better, more comprehensive screening. resulting in fewer false negatives. A study into how the MNA-SF can be efficiently implemented in primary care, whether or not multidisciplinary, could provide an added value.