|
|
Effects of GH Replacement in Adult GH Deficiency on Patient-Reported Outcomes and Cognitive Function: A Meta-Analysis« Back to Volume 22, Issue 2, June 2006 - Table of Contents Arwert and colleagues conducted a meta-analysis to evaluate the effects of human growth hormone (hGH) substitution on patient-reported outcomes (PROs) (ie, quality of life, health status, and well-being) and cognitive function among adults with GH deficiency (GHD). To be eligible for inclusion in the meta-analysis, the study had to provide quantitative data obtained through either a placebo-controlled or cross-over/parallel or open clinical trial. PROs had to be measured by questionnaire and neuropsychological tests for assessing cognitive functioning. A review of PUBMED and Picarta databases (1985 to January 2004) yielded 15 qualifying studies involving 830 patients with 3 to 50 month follow-up. Effect sizes (d-values) were calculated, averaged and pooled for each study. Effect size d = 0.2 is conventionally defined as a small effect, d = 0.5 as a medium effect and d = 0.8 as a large effect. PROs were measured using the Nottingham Health Profile, the Psychological Well Being Schedule, the Hopkins Symptom Checklist, the Profile of Mood States, the State-Trait Anxiety Inventory, and the Quality of Life Assessment of Growth Hormone Deficiency in Adults. A variety of cognitive tests were used to measure short- and long-term memory, comprehension, vocabulary, verbal fluency and non-verbal skills, but these tests were only carried out in 4 of the 15 studies (85 patients, 6–24 month follow-up). Fifteen studies met inclusion criteria and 4 included data on the cognitive effects of hGH treatment in patients with GHD. After 3 months of hGH substitution, PROs in GHD patients exhibited large effect size improvements (5 studies; d = 0.81, 0.32–1.30 [confidence interval, [CI]). However, statistically significant heterogeneity in effect sizes across studies was observed, suggesting the potential influence of moderator variables (eg, participant age, diagnosis, age at onset). Ten studies reported PRO data obtained at 6 months: the effect size was medium (d = 0.55, 0.31–0.79 [CI]) and smaller than after 3 months. Here too, the distribution of d’s was not homogeneous. The effect of 12 months of hGH therapy on PROs was based on d’s from 7 studies: PROs were still increased, but the effect size was smaller than after 3 and 6 months (d = 0.29, 0.11–0.47[CI]). In summary, these results indicate that the longer the hGH substitution therapy lasts, the smaller the effect on PROs (Figure). Five studies included a placebo-treated control group with PROs measured at 6 months: the difference between hGH and placebo-treated groups was not statistically significant (p = 0.55) and the d was almost zero (d = —0.075, —0.32–0.17 [CI]). Across all durations of treatment (median = 6 months; range 1–24 months; 8 studies), the effect was not better for hGH than for placebo (p = 0.85) (d = —0.03, —0.30–0.24 [CI]). In contrast to the effects on PROs, a statistically significant effect of hGH replacement was not observed on cognitive function tests (4 studies). Only 2 studies were placebo-controlled; too few to aggregate findings. Effect sizes (d) of different hGH treatment durations on patient reported outcomes. The authors concluded
that the frequently observed improvement in PROs in open studies may be attributed
to factors other than hGH. These potentially include the attention and care
the patients are receiving and the justification of effort inherent in the
hGH treatment regimen. In regards to the absence of beneficial cognitive functioning,
the authors noted that it is possible that hGH substitution improves some specific
cognitive fields, but pooling all of the different types of cognitive tests
may mask specific effects. Editor’s CommentMeta-analysis involves merging data from multiple studies into single estimates of treatment effect. In theory, this process should enhance the accuracy of estimates of effect size. Of course, the results of a meta-analysis depend on selection criteria for the studies included. An implicit assumption inherent in meta-analyses is that the differences among studies are primarily due to chance. Of course, this is unlikely to be the entire story: instead, differences in the direction or effect size may be attributable to differences in treatments, populations, outcome measures, study design, and study quality. Sonksen and McGauley1 took umbrage with the Arwert report in a comment entitled “Lies, damn lies and statistics.” They point out that at least one apparently eligible study, authored by McGauley2 and which demonstrated a benefit of hGH in PROs employing a double-blind placebo-controlled research design was not included in the meta-analysis. However, had the McGauley study been included, it may have changed the effect size estimate to only a small degree considering the pattern of findings from the other studies included in the meta-analysis. If treating physicians are justifying treatment of adults with GH deficiency, in part, based on anticipated improvements in health-related quality of life outcomes, then further studies of the efficacy of hGH replacement are warranted. The U.S. Food and Drug Administration (FDA) recently posted guidelines to industry regarding the use of patient-reported outcome measures to support labeling claims.3 The FDA document serves as an excellent roadmap for investigators considering research on the benefits of therapeutic agents on quality of life outcomes. However, until such studies warrant changes in labeling, clinicians would be wise to avoid claims of improved health-related quality of life or neurocognitive function when informing patients about risks and benefits of hGH replacement in adulthood. David E. Sandberg, PhD References - (linked to
|
Volume 26 Number 1
September 2010
www.GGHjournal.com
ISSN 1932-9032
)