Chapter 3: Methods for Clinical Research in Constipation: Therapeutic Comparisons

Trials design in palliative care populations

Very few trials of laxative therapies have been performed in palliative care populations. However, it is instructive to compare and contrast the design, outcome measures, dropout rates, and impact on the key findings of those studies that are available.

Table 8.3 Key Studies of Constipation in Palliative Care Populations
Author(s)	Ramesh et al., (1998)	Agra et al., (1998)	Sykes (1999)
Design	Randomized, non-blinded, parallel group design in cancer patients receiving morphine comparing: senna with Ayurvedic laxative preparation called Misrakasneham	Randomized, parallel group, unblinded design in a palliative care population comparing: senna and lactulose	Randomized, crossover design in hospice cancer population comparing: senna + lactulose with danthron + poloxamer
Outcome measures	4-point scale based on the subjects’ level of satisfaction with their bowel actions	# of defecation days divided by # of days in study; plus # of defecation-free 72-hour periods (investigators considered it unethical for patients to go any longer before offering additional intervention)	10-point discrete response scale measured subjects’ level of satisfaction with their bowel actions
Duration	14 days	Variable; up to 34 days	14 days
Dosing	3-level dose titration available according to response after minimum 2-day interval; allowed 4-fold increase in Misrakasneham, and 3-fold increase in senna	4-fold increase in laxative dose permitted; when ceiling-dose of first agent reached, second agent added and dose titrated in turn	4-fold titration range available for each combination
Drop-out rate	28% drop-out rate; drop-outs uneven between groups	18% within first 4 days; 49% failed to complete entire course	56% drop-out rate in 14-day trial period
Adverse effects	Adverse effects of both preparations compared	Adverse effects of both preparations compared	Almost equal numbers of patients (15 for LS, 14 for CD) preferred each laxative preparation; Reports of adverse effects were surprisingly rare. Two patients complained of perianal soreness and burning with codanthramer
Financial cost	Strongly favored Ayurvedic treatment	Senna significantly cheaper than lactulose	At published prices (British National Formulary) the combination of senna and lactulose used in this study was over 40% cheaper than the same, less effective, volume of codanthramer
Key and findings	Despite starting with 25 patients in each arm, the study design and drop-out rate resulted in the trial producing no statistically significant results	37% of patients required both agents for satisfactory bowel function; No significant difference in efficacy between therapies	Significant level of patient-defined "normal bowel function" following use of lactulose + senna; Significant worsening of constipation when danthron + poloxamer followed lactulose + senna, and a non-significant improvement in constipation when it preceded them

It is instructive to consider how trial design impacts the results of studies of constipation in palliative care populations.

In Ramesh (1998), despite starting with 25 patients in each arm of the study, there were no statistically significant results gained because of the dropout rate that was unevenly divided between the two groups.

Agra (1998) illustrates some of the difficulties of research in a terminally ill population, in that only 91 out of 389 potentially eligible subjects could actually be entered into the trial period. Despite this high degree of selection, the dropout rate was 18% within the first four days. However, the design of the trial allowed for variations in the length of the study period, which was a wise precaution as a further 49% of the group failed to complete the entire course. Despite these difficulties, significant results were still obtained.

In the author’s own study (Sykes, 1999), the trial design anticipated the significant dropout rate and allowed for results with statistical significance. Given the standard deviation of the stool frequencies, a group size of 13 was sufficient to deliver a study power of 80% by employing a crossover design. If a parallel group design had been employed, 50 subjects per arm would have been required to achieve the same study power. As most of the drop-outs occurred in the first week rather than the second, the trial would have failed to achieve a statistically significant results if it had used a parallel group design.

Page 27 of 36
Previous Page	Next Page
	Next Section