News · Nutrition & Diet
Feeding malnourished older inpatients properly tracked lower death rates and shorter stays
Hospital dietitians are stretched across every ward. Tracking 17,057 inpatients averaging 78 found the gains concentrated entirely in the patients who arrived already malnourished.
- Among malnourished patients, deaths in hospital were about half as common.
- Six-month deaths and length of stay both moved the same way.
- Patients who were not malnourished showed no benefit at all.
- More of the framework delivered tracked shorter stays.
- The results did not survive statistical correction for testing several outcomes.
Hospital food is a joke with a serious problem underneath it. A frail eighty-year-old admitted with pneumonia may be eating almost nothing for a week, and the chart will record the antibiotic rather than the untouched tray.
Dietitians exist to catch this and there are never enough of them. So the practical question is not whether nutrition matters but where to point a small team.
What the ward nutrition framework changed
Writing in JAMA Network Open, researchers examined what happened when wards adopted a structured nutrition routine: screen everyone on admission with a short questionnaire, act on the score, and follow through.
Among 17 057 inpatients included in the study, mean age 78, they compared outcomes, splitting them by their screening result into malnourished, at risk, and normally nourished.
Previously the case for this kind of framework rested on trials of supplements rather than on whether reorganizing a ward changes what happens to patients.
Why only malnourished inpatients benefited
In the overall group, nothing. Across all 17,000 patients, the framework was not associated with lower mortality or shorter stays.
Among the patients who arrived malnourished, in-hospital deaths were roughly half as common. Deaths by six months were lower as well, and stays were shorter.
That contrast is the finding. A change to how a ward handles nutrition did nothing for people whose nutrition was already adequate, and appeared to do a great deal for the people whose was not.
Why the inpatient mortality findings lost significance
These results did not survive correction for testing several outcomes at once.
That deserves stating plainly rather than burying. When you test three outcomes, one of them clearing the usual threshold by chance is unsurprising, and a correction raises the bar accordingly. On the corrected standard, none of these findings clears it.
What argues against dismissing them is the pattern rather than any single number. Three separate outcomes moved the same direction. A dose-response appeared for length of stay, so wards delivering more of the framework saw shorter stays. And a Bayesian analysis put the probability of genuine benefit above 96% for every outcome.
None of that is proof. It is the difference between a lone borderline result and a coherent one.
What a before-and-after hospital comparison cannot do
This compares a period after implementation against a period before it, which means everything else that changed over that time is mixed in. Hospital care does not hold still.
Nobody was randomized, and wards that adopted the framework more thoroughly may differ in other ways that also help patients. The dose-response is suggestive and it is not immune to that.
The authors are explicit that confirmation in multicenter prospective studies is needed, which is the right conclusion from what they have.
How to target nutrition care at the patients who need it
Good nutrition is important, no matter what your age, and in hospital the obstacle is rarely disagreement about that. It is that nobody is assigned to notice.
The practical value here is in targeting. If the benefit is concentrated in patients who screen as malnourished, then a stretched dietetics service has a defensible rule for where to spend its hours, and a two-minute questionnaire on admission is what identifies them.
People also ask
What did the study find?
Among 17,057 inpatients (mean age 78.1 years; 53.2% male), in-hospital mortality in the malnourished stratum was lower during implementation (odds ratio 0.45; 95% CI, 0.22-0.91; P = .03), alongside lower six-month mortality (0.59; 0.35-0.98; P = .04) and shorter length of stay (exponentiated coefficient 0.87; 0.77-0.98; P = .03). After Benjamini-Hochberg adjustment these were no longer significant, though the Bayesian posterior probability that the odds ratio was below 1 was at least 96% for all outcomes.
What is a nutrition clinical framework?
A structured ward routine for identifying malnourished patients and acting on it: screening on admission, a plan, and follow-through on what the patient actually eats, rather than leaving nutrition to be noticed.
How is malnutrition identified here?
By a short validated questionnaire scored out of 14. Seven or below counts as malnourished, 8 to 11 as at risk, and 12 or above as normal. It takes minutes and does not require a blood test.
What does it mean that the findings did not survive adjustment?
When several outcomes are tested at once, some will look significant by chance, and a correction raises the bar to account for that. These results cleared the ordinary threshold but not the corrected one, which makes them suggestive rather than established.
Why report them at all then?
Because three different outcomes moved the same way, a dose-response pattern appeared for length of stay, and a Bayesian analysis put the probability of benefit above 96%. That combination is harder to dismiss than any single borderline result.
Why did well-nourished patients show nothing?
Because there was nothing to correct. A framework that improves nutritional care can only help people whose nutrition is inadequate, and finding no effect in the rest is what you would expect if the effect is real.
What does this mean for a patient or family?
Mainly that nutrition during a hospital stay is worth asking about for a frail older relative. Clinical decisions belong to the treating team. This is general information rather than medical advice.