This study provides support for the validity and user-friendliness of the MEONF-II and the NRS 2002. Perceived user-friendliness of the MEONF-II was somewhat better than that of the NRS 2002. It was indicated, although not significantly, that the MEONF-II had better sensitivity than the NRS 2002 in comparison to the MNA.
The MNA has commonly been used as the comparator, or gold standard nutritional screening tool (e.g., [7
]). MNA captures patients at risk also in an early stage so that preventive actions can be taken [12
]. The NRS 2002 differs somewhat in focus in that its goal is to identify patients that are most likely to benefit from nutritional interventions [4
]. It is therefore possible that the NRS 2002 is less suited for detecting patients in need for preventive actions than the MNA is. The intention of the MEONF-II, on the other hand is both to identify patients needing preventive nutritional interventions and those needing nutritional treatment. Furthermore, the MNA was developed for people ≥65 years [12
], and the use of it as gold standard among younger patients can be questioned. However, we did not find any relevant differences between results from analyses of the full sample and when excluding the younger subsample. Thus, whether the MNA can be considered an appropriate gold standard or not depends on the intention with the comparator and possibly the age of the patients in the sample. In any case, people at risk for undernutrition need further assessments and no instrument can alone capture all aspects influencing the eating situation and the nutritional intake.
Cut-off scores used in the different instruments affect the results. The MNA cut-offs have been defined based on serum albumin values, a predictor of morbidity and mortality in elderly people [13
]. This would suggest that any comparison with the MNA may only indicate whether low albumin levels can be detected. However, the MNA has also been validated against more extensive assessments of nutritional status (including, e.g., additional biomarkers and dietary parameters) [6
]. The cut-off scores for MEONF-II have been defined based on clinical reasoning and confirmed by ROC analysis against the MNA classification [8
], and the NRS 2002 cut-off was based on findings from randomized controlled trials regarding the effect of nutritional intervention [4
]. However, the classification of patients in that study was done retrospectively and the authors were not blinded to outcome (usually artificial nutrition) when estimating the degree of undernutrition and severity of disease [4
]. The different intentions of these instruments, the way they have been developed and compared with other measures affect the prevalence findings. For instance, in this study the NRS 2002 identified a significantly lower percentage (29%) of patients as at risk than the MEONF-II (45%) and MNA did (60%). Similarly, in another study [15
] the MNA identified 70% of patients as at risk or malnourished while the NRS 2002 identified 40%. It therefore appears that the MEONF-II does not identify patients at risk as early as the MNA, and not as late as the NRS 2002. In clinical practice such differences will have consequences for preventive and treatment actions. Further on, a majority of those being undernourished according to MNA were correctly classified as at high risk by MEONF-II (13 out of 18 patients) or at risk by NRS 2002 (12 out of 18 patients). Anyhow, it should be remembered that the main purpose with screening is to identify people at risk and not to decide whether it is a low or high risk and that any case being at risk needs a more detailed assessment. In addition, efforts are needed to develop a clear vocabulary and uniform definitions of risk (low/high) and manifest undernutrition.
The MEONF-II showed a 68% concordance with the MNA, which is lower than that observed in a previous study (82%) [8
]. One explanation to the difference in accuracy could be that in the previous study [8
], the assessment procedures were reviewed individually with the nurse assessors, whereas it was conducted as a group session in this study. However, the accuracy and sensitivity of the NRS 2002 found here (55% and 37%) are similar to those in previous studies of this instrument in relation to the MNA (52% and 39%) [15
The MEONF-II is a screening tool designed to detect risk of undernutrition, not only those with manifest undernutrition. As such, it is reasonable for sensitivity to be given priority at the cost of specificity since over-identification is preferable to under-identification, given that positive screening results are followed by in-depth assessment [19
]. In this respect the NRS 2002 appears less well suited, since its sensitivity was lower (37%) compared to that of MEONF-II (61%). However, as the associated 95% CIs overlapped, additional studies in larger samples are needed before any firm conclusions can be drawn.
The MEONF-II demonstrated good user-friendliness in terms of time to complete, ease of understanding of items, as well as ease of completion. In these respects, our observations suggest that MEONF-II compares favourably to the NRS 2002. One reason for this may be that it helps nurses identifying problems and intervene directly, either themselves or by involving other professionals. It should be noticed that time consumption was low when using MEONF-II, despite the fact that this assessment was done before NRS 2002 and MNA. It could otherwise be expected that the time needed would be lower for tools used as second and third since several items are shared between the tools. One should, however, be careful in the interpretation of these findings since it was only four nurses that rated user-friendliness and there may be a learning curve for each of these nurses, affecting rating of user-friendliness and time needed for completing forms. Anyhow, user-friendliness is of fundamental importance for successful clinical implementation of nutritional screening tools. As most screenings are carried out by nurses, their perspective in this respect must be taken into account.