Socio-economic, geographic and occupational characteristics
Of the 316 experts invited, 93 participated in the first Delphi round. The most common age group was 40-49 years old (32%, yeah = 30), and the distribution of other age groups was similar, except for the underrepresentation of the 20-29 year old population (3%, yeah= 3) (Table 1). Most of the participants were women (77%, yeah= 72), Europe (72%, yeah= 67), followed by Asia (14%, yeahParticipants came from 27 countries, with the majority coming from the Netherlands, Turkey, the UK and Portugal. Two-thirds (66%) yeah62% of participants were geriatricians or researchers, and 16% (yeahMore than half of the participants were dieticians or nutritionists (55%, yeah= 51), 36% (yeah= 33) in the community, and 10% (yeah= 9) are receiving long-term care, while nearly half (48%, yeah= 45) reported that the primary follow-up period for nutritional interventions was less than 12 weeks. The response rate for the second Delphi survey was 77.4% (yeah= 72). The distribution of traits is similar in the first and second rounds, and importantly, Europe is still over-represented in the latter (75%, yeah= 54) (Table 1).
Delphi Survey Round 1
71% agreed with the exclusion of nine non-important outcomes (yeah= 66) participants, 20% (yeah= 19) considered that physical activity should be assessed in round 2, which was done. For all other outcomes, only ≤4 participants did not agree to be excluded.
In this round, more than 75% of participants rated malnutrition status (88%), dietary intake (83%), weight or BMI (75%), muscle strength (82%), and functional performance (85%) as important, and more than 60% rated the PROs functional limitations (72%), quality of life (80%), and intervention acceptability or adherence (79%) as important for inclusion in the COS ( Fig. 1A ; Table S2 ).
a The difference between the first-round participants’ ratings and the inclusion thresholds for selected outcomes (60% for PROs and 75% for non-PROs) (yeah= 93) and B Delphi survey round 2 (yeahAdverse events were excluded, and physical activity was included in the second round of the survey. C Final consensus meeting: Percentage of participants who agreed to include these outcomes in the malnutrition COS (yeah= 15). Results showing 70% or higher concordance were included in the COS. 
and 
Bars denote outcomes above and below the consensus threshold according to the protocol, respectively.*Outcomes voted together for exclusion include body circumference, skinfold, mortality, healthcare utilization, healthcare costs, comorbidities, health status, severity of dysphagia, fatigue, frailty, self-perceived health, pain, cognitive status, depression, anxiety, sleep disorders, self-esteem, hydration status, feeding behavior, energy requirements, blood markers, nitrogen teller, participation in social roles and activities, peak expiratory flow, bone health, falls, and physical activity.It is worth noting that weight loss can be calculated by assessing weight at baseline and follow-up.BMI Body Mass Index, COS Core Outcome Set.
None of the findings were judged to be so unimportant that they needed to be re-evaluated in a subsequent Delphi survey round (more than 75% unimportant and less than 15% important for non-PROs; more than 60% unimportant and less than 15% important for PROs). Therefore, because it was unclear which findings to exclude, participants were asked to re-evaluate these findings in a second Delphi survey round.
The evaluation of the results of the first round of the Delphi survey was carried out in the setting (community yeah= 33, Hospital yeah= 51 and long-term care yeah= 9) (Table S3) and follow-up period (<12 weeks yeah= 45, 12 weeks or more yeah= 48) (Table S3). However, there were some important exceptions worth mentioning, although not enough to justify different COS by setting. For example, in the hospital setting, more than 75% (77%) of participants rated mortality as an important outcome to include in the COS, whereas in the community setting, fewer than half (49%) did so. Similarly, 71% of participants in the hospital setting rated complications as important, whereas only 39% in the community setting did so. Very few participants listed long-term care as their primary setting (yeah= 9), but the severity of dysphagia (yeah= 8) and hydration status (yeah= 7) was rated as more important than other settings (Table S3).
Delphi Survey Round 2
Adverse events were excluded from the second round of reassessment because the steering group determined that reporting of adverse events is already mandatory in any trial and therefore did not need to be included in a specific COS. Furthermore, adherence to the intervention was excluded from the acceptable range of outcomes because it is too different in concept and for the same reasons.
Nearly all participants (99%) in the second round agreed that malnutrition status, weight or BMI, functional performance, dietary intake, muscle strength, functional limitations, quality of life, and acceptability of the intervention should be included in the COS. One participant disagreed with including acceptability of the intervention and quality of life (Figure 1B; Table S3).
Additionally, over 75% of participants re-rated muscle mass (82%) and frailty (79%) as important for inclusion in the COS, and over 60% re-rated appetite (64%) as an important PRO for inclusion in the COS.
Verification by PPI representative
Five PPI delegates from the Netherlands, Turkey and Portugal (3 adults aged 80 years or older: 1 malnourished, 2 at high risk, 2 informal carers) considered health, good appearance, walking unaided and fatigue, weight regain, strength, physical ability, ability to play sports and memory as outcomes important to them. The steering group concluded that these outcomes largely mirrored the results of the two Delphi rounds, except for memory and good appearance, whose cognitive status was rated low in both Delphi rounds. It was discussed that good appearance may have some overlap with self-perceived health (rated low), self-esteem (rated low) and quality of life (included in the rating) and that this type of outcome may become even more important in the future and that specifically designed studies are needed.
PPI representatives also generally agreed with the results obtained in the two Delphi rounds, except for functional performance, which had five neutral votes (one malnutrition status, one muscle mass, two frailty, and one acceptability of the intervention) and one negative vote.
Final Agreement Conference
The final consensus meeting took place online on 27 July 2023 and was attended by 15 people, including one chair (who abstained). Attendees included two PPI representatives (one of whom joined midway through the meeting), six steering group members, five participants from both Delphi rounds (selected for their involvement in health policy or status as dietitians), and two external guests from the medical nutrition industry.
Participants were asked whether they agreed to exclude undetermined outcomes (i.e., body circumference, skinfold, mortality, healthcare utilization, healthcare costs, complications, health status, severity of dysphagia, fatigue, frailty, self-rated health status, pain, cognitive status, depression, anxiety, sleep disorders, self-esteem, hydration status, eating behavior, energy requirements, blood markers, nitrogen balance, participation in social roles and activities, peak expiratory flow, bone health, falls and physical activity) from both Delphi rounds. No consensus was reached as less than 70% (64%) of participants agreed to the exclusion of these outcomes. Participants who did not agree to the exclusions agreed to the exclusion of healthcare costs, self-rated health status, hydration status, physical activity, body circumference, adverse events, complications or mortality. A group discussion was then conducted. Although healthcare costs were considered particularly important, participants decided that it should not be mandatory for small trials and should be recommended for larger trials where a cost-effectiveness analysis might be performed. Self-rated health was considered to be at least partially reflected in quality of life and therefore did not require further inclusion in the COS. Participants considered hydration status to be less important when considering the impact of malnutrition treatment, but acknowledged that nutritional interventions are not just about food and nutrients. The inclusion of physical activity in the COS was not supported because physical activity was under-rated in the second Delphi round and participants in the consensus conference considered that individuals could be physically active or inactive regardless of nutritional status. Body circumference was mentioned as a simple and viable measure of muscle mass. However, this comment was howOutcomes need to be measured, but this is not relevant to this phase of COS development.After group discussion, 80% of participants (out of 15, as both PPI representatives were present) voted again to remove all pending outcomes from the COS (Figure 1C).
All outcomes that reached consensus in at least one of the Delphi rounds were voted for inclusion in the final COS, except for frailty, which did not reach ≥70% consensus in the final vote because the group considered that frailty elements largely overlapped with other COS outcomes ( Figure 1C ).
