Clinical Drug Experience Knowledgebase

Effects of Transdermal Testosterone and/or Monthly Vitamin D on Fall Risk in Pre--frail Hypogonadal Seniors. Double Blind, 2x2 Factorial, Randomized Placebo--Controlled Clinical Trial

Background:

One of the challenges of demographic change is the enormous personal and economic burden of falls and their consequences in the senior population. Thirty percent of adults 65 years or older and 40 - 50% of those age 80 years or older fall once per year. Serious injuries occur in 10 - 15% of falls, resulting in fractures in 5% and a hip fracture in 1-2% of cases. As an independent determinant of functional decline, falls lead to 40% of all nursing home admissions.

Several studies have estimated the prevalence of low testosterone levels (hypogonadism) in older men to be about 20% at age 60. This proportion increased in the Baltimore Longitudinal Study of Aging with each decade (30% at age 70 and 50% at age 80). Low testosterone levels are associated with decreased strength, impaired physical performance, and an increased risk of falls, and, thus, low testosterone may contribute to the frailty syndrome in older men. Current studies have focused on healthy, community dwelling men with a high level of function. Our trial aims to enrol pre-frail senior men at risk of losing their autonomy. Further, in this population and also relevant for muscle health and falling is the high prevalence of vitamin D deficiency.

Dose Rationale:

The investigators chose 75 mg testosterone/day as the minimally sufficient dose to shift over 90% of participants to total testosterone levels of at least 16.86 nmol/l (lower end of the third quartile in the pilot trial where fall reduction was significant -- Bischoff-Ferrari et al.; Osteoporosis International 2008). Additionally, this dose has been chosen for its cardiovascular safety as summarized in a 2013 review.

The investigators chose 24000 IU/month of vitamin D (equivalent to 800 IU vitamin D per day) as the current recommended dose for vitamin D and its evidence for fall reduction. To allow for the standard of care and given that 50% of participants will not be randomized to vitamin D supplementation - all participants are allowed to take 800 IU vitamin D per day in addition to the study medication.

Randomization:

Participants will be randomized at the baseline clinical visit using a computer-based randomization procedure. Willing and eligible participants will be randomized to one of the four treatment groups stratified by age (2 groups: 75 - 79 and 80+). Within each stratum, treatment assignments will be generated in blocks of 8 individuals, with 2 individuals in each of the 4 treatment groups.

Blinding:

Study medications will be labelled with a unique randomization code at the respective companies. The codes will be inaccessible until the data set is frozen and the study is unblinded. All study packages will look identical, except for the randomization codes on the labels.

Follow-up:

Follow-up is 12 months with 3 clinical (Baseline, 6 and 12 months) and 4 (2, 4, 8 and 10 months) phone call visits.

Determination of Sample Size:

The sample size of 168 senior participants was chosen to have sufficient power for the primary endpoint (odds of falling) and the effect of testosterone treatment. Under these assumptions and assuming an additive effect of the 2 interventions (testosterone and vitamin D), as expected from the investigators pilot study, they have 89% power for the primary endpoint "number of participants sustaining at least 1 fall". This computation assumes that the difference in proportions of fallers is -0.24 (specifically, 0.38 versus 0.62 - based on the investigators pilot study). If 17% of participants would contribute no data (total n = 140), the investigators would still have 82% power to detect this difference.

Rate of falls: under the same assumptions, and with the proposed sample size of 84 and 84 for the two groups, the study will have power of at least 90.6% to yield a statistically significant result for the rate of falls. This computation assumes that the mean difference in rate of falls is -1.0 (corresponding to means of 0.9 versus 1.9) and the common within-group standard deviation is 2.0 (based on standard deviation estimates of 2.2 and 1.7).

Planned Analyses:

Primary analysis will be conducted according to the intention to treat (ITT) principle. Every study participant who was randomized and received the study treatment package at baseline clinical visit will be a part of the primary analysis population. In addition to the primary ITT analyses, the same data will be analyzed in participants who demonstrated acceptable adherence to the study protocol. The per protocol analyses will be treated as sensitivity analyses. The per protocol analysis population will include the participants who: Attended at least 1 out of the 2 follow-up clinical visits; Participated in at least 2 of the 4 follow-up phone calls; Took at least 80% of the study medication(s) as confirmed by count of used bottles for drinking solution and count and weight of used transdermal gel containers; did not take additional vitamin D and calcium supplements in doses exceeding those stipulated by the exclusion criteria.

All assessments of treatment effects in the efficacy analyses will be based on the intent-to-treat principle. All endpoint variables will be summarized by treatment group. N, Mean, Standard Deviation, Minimum and Maximum will summarize continuous endpoint variables and Number and Percentage will summarize categorical endpoint variables. All endpoints will be tested at the 2-sided at 5% significance level (α = 0.05, 2-sided). Two types of models will be run for each analysis. Minimally adjusted models will include age, body mass index, and baseline serum levels of total testosterone and 25-hydroxyvitamin D; fully adjusted models in addition to the parameters included in the minimally adjusted models will also include number of comorbidities, cognitive function, and additional potential confounders if baseline analysis justifies their inclusion. Each of the models will initially check for possible interaction between the study treatments. If no such interaction in found, the interaction term will be dropped and the simplified models will be used to estimate separate effects of the study treatments.