had travelled through several time zones in the preceding 9 days.
Of the other 8 trials, four examined a 'pre+post', and four a 'post' regimen. All were performed in
civilian travellers.
In all nine trials the clock time at which melatonin was taken after arrival at the destination remained
the same each day: the time of administration was not changed 1-3 hours per day, to take account of
the way in which the body clock is expected to adjust (Lewy 95).
Two trials (87 Arendt, 91 Nickelsen) included assays of cortisol and of melatonin to measure
circadian phase.
The four trials of the 'post' regimen, all in travellers, differed in size (92 Nickelsen n=36; 97 Spitzer,
n=257; 98 Suhner a, n=234; 98 Suhner b, n=137) and in design.
The Nickelsen study focused more on endogenous cortisol and melatonin rhythmicity than on
symptoms of jet-lag. Suhner's first trial (98a) compared two doses of fast-release melatonin with an
intermediate dose of slow-release melatonin and placebo after an eastward flight from Switzerland to
Asia, or America to Switzerland. The volunteers completed sleep logs, a symptoms questionnaire
and the Profile of Mood States (POMS) daily for 3 days before and 4 days after the flight; subjective
sleepiness was assessed three times a day. In Suhner's second trial (98b) melatonin was of subsidiary
interest: the prime focus was on zolpidem with which melatonin was compared, and the results for
melatonin and placebo are not reported in detail. The Spitzer trial compared three different regimens
of melatonin with placebo in Norwegians (mainly physicians) who were returning to Oslo after
having visited New York for 5 days.
Methodological quality
See:
Table of included studies
The design and performance of all ten trials seem to have satisfactorily minimised selection bias,
performance bias, attrition bias and detection bias (Cochrane Handbook 99). All except that of
Claustrat are described as double-blind, but none of the reports includes a statement on allocation
concealment, on how closely alike in appearance, etc, the test treatments were, or whether
participants could ever examine or see different treatments side by side.
None of the reports state what the participants were told about the trial they were entering, and what
effects they would have been led to expect. The prior expectations of the participants could well
have influenced the effects and symptoms that they experienced and reported.
None of the trial reports give details of the source of the melatonin used and most do not state the
pharmaceutical form used.
Nickelsen's trial was underpowered because it was not possible to use a cross-over design. The
authors also note that they could not control how closely their volunteers followed the protocol in
respect of regular bedtimes, abstention from alcohol, etc.
In Spitzer's trial the participants had come from Norway to New York and after 5 days would not
have fully adjusted to New York time before they flew back. They would therefore have been
expected to suffer less jet-lag than thoroughly adapted transatlantic travellers, so that a reduction of
jet-lag by melatonin would have been more difficult to detect in this trial. For this reason the results
of this trial are not considered in comparisons (4) and (5).
The reporting of adverse events in the trials is mostly rudimentary and inadequate. In some the
participants were asked specifically, in most only spontaneously mentioned events are noted. The
number of reported adverse events is often not explicitly related to the number of individuals who
experienced them.
Other reports of adverse events span a huge range in quality, from the detailed and unambiguous to
many that are fragmentary and uninterpretable, with most at the latter end of the range.
Results
!
List of comparisons
!
Additional tables
Comparison (1): melatonin v. placebo
Table 01
In nine of the ten trials melatonin reduced the symptoms of jet-lag. In one trial (Spitzer) no effects
were found, but this can be attributed to a faulty design (see above). For 5 trials in travellers (87 &
88 Arendt, 89 Petrie, 91 Nickelsen, 92 Claustrat), totalling 237 flights, a global visual analogue jet-
lag score is reported. We have converted these scores to a single scale from 0 to 100. Table 1 shows
the results: the weighted mean score after after melatonin was 25, after placebo 48, (excluding the 18
westward flights in Nickelsen's trial). The meta-analysis, which allows for the differing variances in
the trials, shows a bigger difference in score of 37.3 (95% CI 39.8 - 34.9). This is highly significant,
both statistically and practically.
In the only two trials that reported results for individuals (and not merely group means), 16 out of 24
people (67%) given placebo experienced jet-lag after an eastward transatlantic flight (87 Arendt, 92
Claustrat), while only 4 out of 22 (18%) did so after 5mg melatonin. On this basis, one of every two
people taking melatonin would benefit. The group means reported in the other trials are consistent
with such an estimate.
Various components or aspects of the jet-lag syndrome were assessed in the trials, but the results
cannot be combined because the methods of measurement and reporting differed.
SLEEPINESS was rated and reported on in three trials: Claustrat found significantly less morning
tiredness and less evening sleepiness for up to 6 days with melatonin than with placebo; 98 Suhner
(a) found melatonin takers less tired from the second day on; in 93 Petrie cabin crew taking late
melatonin were less sleepy than those on placebo or 'early+late' melatonin.
SLEEP LATENCY was reported by 87 Arendt as significantly shorter with melatonin than with
placebo for 6 days after the flight, and SLEEP QUALITY likewise; Suhner (a) found the differences
in latency and quality greatest on day 2; the Claustrat trial found no clear difference in latency, but
used a coarse and insensitive scale.
The Profile of Mood States used by 89 Petrie showed that in comparison with the placebo group,
melatonin increased 'vigour/ activity' and lessened fatigue. 98 Suhner (a) noted that fatigue scores
were similar for 5mg and 0.5mg melatonin, both less than for placebo. 89 Petrie reports three useful
subjective estimates of RECOVERY: how many days it took a) for the sleep pattern to return to
normal, b) for energy likewise, c) for daytime tiredness to disappear. All three became normal sooner
with melatonin.
SYMPTOMS are reported in several trials, but there is no certain way of deciding whether a
symptom is due to jet-lag, to melatonin or to something else. When a symptom occurs more often
after melatonin than after placebo, causation in an individual case remains in doubt.
Comparison (2): 'post' regimen v. 'pre+post' regimen
The 93 Petrie trial in airline cabin crew was the only one that directly compared a 'pre+post' with a
'post' regimen, as well as with placebo. In the 'pre+post' group overall recovery after the flight was
worse than in the placebo group, whereas the 'post' melatonin group had less jet-lag and sleep
disturbance than the placebo group. In this trial however the circadian rhythms in the participants
were so disordered that the timing of the pre-flight doses must have been very variable in their
circadian phase.