Rybarczyk, B., Lopez, M., Benson, R., Alsten, C., & Stepanski, E. (2002). Efficacy of two behavioural treatment programs for comorbid geriatric insomnia. Psychology and Aging, 17(2), 288-298.
Rybarczyk et al. (2002) found that the literature addressing geriatric insomnia provides evidence of the efficacy of several behavioural treatments, but that most research with this population has excluded those with comorbid medical illness. Thirty-eight geriatric patients with comorbid insomnia and medical illness participated in this randomised controlled trial, which allocated them to either cognitive-behavioural treatment (CBT), home-based audio relaxation treatment (HART), or delayed-treatment control conditions using stratified random allocation.
Seven self-rate or self-report and one actigraphy measures assessed between- and within-group differences at pre-treatment, post-treatment, and 4-month follow-up. The CBT included weekly group sessions with behavioural and cognitive components, relaxation training, and sleep hygiene education. The HART participants received audiotape recordings with four relaxation and sleep-training programmes and a guidebook with behavioural components. Control participants completed the same interval assessments as the treatment groups and were offered HART at the end of 6 months.
The study found differential improvement between the CBT and control groups at post-treatment and follow-up on five of seven sleep measures and a clinically significant difference for sleep efficiency between the HART and control groups at post-treatment. It concluded that CBT is an effective treatment for comorbid geriatric insomnia. Home-based treatment should, however, still be considered as a less-expensive alternative, as it did show some benefit.
The hypotheses are that behavioural treatments are effective for individuals with comorbid geriatric insomnia and that participants who experienced a significant improvement in sleep would have secondary psychological and quality-of-life benefits. The study did not, however, state which dependent variables correspond with which outcome measure or measures. This and such poor operationalisation of the variables as treatment efficacy limit its construct validity.
The study used a delayed treatment with the control participants, as it is unethical not to treat patients. However, which treatment to offer was determined prior to obtaining the results, which led to the controls receiving the less effective treatment. The researchers also used a patient database to recruit prospective participants seemingly without consent, possibly breaching the privacy and confidentiality ethics.
The study failed to identify the tool it used to establish whether patients met the diagnostic criteria for insomnia, limiting the validity and reliability of its diagnoses. It also failed to document the severity of each patient’s symptoms, which may have confounded both sleep and secondary results if one degree of severity responded better to treatment.
The study measured negative sleep-related cognitions, but reported the results without providing evidence of a relationship between insomnia symptoms and such cognitions. Neither of its hypotheses, furthermore, proposed this relationship. Even though both treatment programmes have key components that focus on cognitive restructuring, the links between this and the research problem are unclear.
The study reported both clinically significant improvements and statistically significant findings, but excluded such standardised outcome descriptions of patient benefit as functioning within normal limits. These are important for future research comparisons and repeatability.
The results might have included a measure of the r value for the scale or scales that were measuring the same dependent variables. This calculation would have indicated whether the correlation was high enough to combine them into a single index, thus making it easier to evaluate the results in accord with the hypotheses.
The scarcity of research on this topic can explain the ambiguity of its construct operationalisation and thereby its paucity of significant results. However, the conclusion validity remains limited and due to the small sample size (n=38), and in turn the limited range of participant characteristics, the results are largely ungeneralisable to the population.