Evidence strengths and limitations

Advice

The main limitations of the 2 trials in people with autonomic diabetic neuropathy (Campbell et al. 1975) and people with Parkinson's disease (Schoffer et al. 2007) are the small study sizes (n=6 and n=17 respectively). Campbell et al. (1975) also focused on disease-orientated outcomes (change in standing blood pressure performed using tilt-table testing), as opposed to patient-orientated outcomes, such as light-headedness, fainting or falls. Only the trial in people with chronic fatigue syndrome (CFS; Rowe et al. 2001) had a reasonable sample size (100 people). This study did assess patient-orientated outcomes; however, these were of more relevance to a population with CFS rather than postural hypotension. Schoffer et al. (2007) did include clinical outcomes of relevance to postural hypotension – the orthostatic domain of the Composite Autonomic Symptom Scale (COMPASS‑OD) and a global impression of change score focusing on orthostatic symptoms. However, Schoffer et al. (2007) did not report any statistical analysis for 2 out of the 3 primary outcome measures.

Two of the trials appeared to have attempted allocation concealment (Schoffer et al. 2007 and Rowe et al. 2001).

The trials by Campbell et al. (1975) and Schoffer et al. (2007) included people with postural hypotension as a result of autonomic diabetic neuropathy and Parkinson's disease, and the trial by Rowe et al. (2001) included people with CFS who had neurally mediated hypotension provoked during tilt testing. Results from these trials may not apply to people with postural hypotension from other causes or conditions.

In Rowe et al. (2001), the largest of the trials, there was a higher level of withdrawals from the fludrocortisone group (26%) compared with placebo (16%). Rowe et al. (2001) reported using an intention-to-treat analysis for reporting of the primary outcome (change in global wellness score of more than 15 points). However, the study authors reported that only 83 of the 100 participants had adequate outcome data for analysis, so it is unclear if the analysis was in fact performed by intention to treat. Complete outcome data for the second tilt-table test that was performed in the 9th week of treatment were not available for 20 people in the fludrocortisone group and 10 people in the placebo group. Campbell et al. (1975) and Schoffer et al. (2007) did not use an intention-to-treat analysis.

Another limitation of the trials was that blood pressure was recorded on a single assessment, and this may miss significant changes (Schoffer et al. 2007).

The RCTs included in this evidence summary were all of short‑term duration and therefore do not provide information about the longer-term safety and efficacy of fludrocortisone.

An observational study by Hussain et al. (1996) did examine safety of fludrocortisone prospectively over an average follow‑up of 12 months. However, the observational nature of this study and the lack of a control arm limit the interpretations that can be drawn.