A plausible biological benefit does not guarantee a favorable clinical outcome. Two randomized trials reported by Takala and colleagues in 1999 illustrate why survival outcomes must take priority over an anabolic hypothesis. Primary study record.
What the trials found
Two parallel, double-blind placebo-controlled studies involved 247 Finnish patients and 285 patients elsewhere in Europe with prolonged critical illness. In-hospital mortality with high-dose growth hormone versus placebo was 39% versus 20% in the Finnish study and 44% versus 18% in the multinational study. Among survivors, intensive-care and hospital stays and ventilation duration were longer with growth hormone. Original report, available through Stanford.
Understanding the absolute differences
Calculated from those rounded rates, the absolute mortality differences are 19 percentage points and 26 percentage points, respectively. These are separate study estimates, not a pooled result or an individual patient’s predicted risk. The calculation is treatment-group mortality minus placebo-group mortality.
A rise from 20% to 39% is not an increase of only 19% relative to the original risk. Absolute and relative changes use different denominators. Keeping both event rates visible communicates the scale without hiding the study population.
Our appraisal
An intervention can improve a laboratory measure while worsening outcomes that matter more to patients. A proposed reduction in tissue breakdown cannot counterbalance demonstrated excess mortality merely because it fits a persuasive mechanism.
The two studies should remain distinguishable. Their populations and mortality rates are not one interchangeable dataset, even though they addressed a related question in parallel. A combined summary should not obscure that the harmful direction appeared in both.
Survivor-only comparisons require additional care. The people who survive are a selected subset after treatment assignment. Interpretation of their length of stay should account for the different mortality experience, rather than treating survival-conditioned groups as unchanged randomized populations. This is our methodological interpretation, not a new outcome reported by the investigators.
What the findings do not establish
These results concern the investigated high-dose strategy in prolonged critical illness. They are not an identical risk estimate for every growth-hormone indication or regimen. Equally, evidence from another setting cannot erase this result. This page does not establish a mechanism for the excess deaths or authenticate any separately supplied research material.
Frequently asked questions
Why include a harmful result in a research library?
A balanced evidence collection must account for findings that contradict the proposed benefit. Omitting them would distort the evidence available to readers.
Does an anabolic mechanism establish recovery benefit?
No. A mechanism provides a rationale to investigate an intervention; functional outcomes and harms still need direct evaluation.
Does this page provide treatment instructions?
No. It is an educational appraisal of a historical study, not medical advice or a dosing guide.
Source and review scope
Takala J and colleagues. Increased mortality associated with growth hormone treatment in critically ill adults. New England Journal of Medicine. 1999;341:785–792. DOI and publisher record. Later correspondence should not be counted as an independent trial.
Prepared with AI assistance; source and editorial checks completed October 1, 2026. No independent medical expert review is claimed. This is an appraisal of the named report, not a systematic review of growth hormone.
Continue with the Research Library and the guide to randomization, masking and control groups.