Temazepam at high altitude reduces periodic breathing without impairing next-day performance: a randomized cross-over double-blind study


Annabel Nickol, 4 Home Farm Cottages, Bletchingdon, Oxfordshire OX5 3DD, UK. Tel./fax: +44-1869-351841; e-mail: annabel@medex.org.uk


The aim of the study was to examine the efficacy and safety of temazepam on nocturnal oxygenation and next-day performance at altitude. A double-blind, randomized, cross-over trial was performed in Thirty-three healthy volunteers. Volunteers took 10 mg of temazepam and placebo in random order on two successive nights soon after arrival at 5000 m, following a 17-day trek from 410 m. Overnight SaO2 and body movements, and next-day reaction time, maintenance of wakefulness and cognition were assessed. Compared with placebo, temazepam resulted in a reduction in periodic breathing from a median (range) of 16 (0–81.3)% of the night to 9.4 (0–79.6)% (P = 0.016, Wilcoxon's signed-rank test), associated with a small but significant decrease in mean nocturnal SaO2 from 78 (65–84)% to 76 (64–83)% (P = 0.013). There was no change in sleep latency (P = 0.40) or restlessness (P = 0.30). Temazepam had no adverse effect on next-day reaction time [241 (201–380) ms postplacebo and 242 (204–386) ms post-temazepam], maintenance of wakefulness (seven trekkers failed to maintain 40 min of wakefulness postplacebo, and four post-temazepam), cognition or acute mountain sickness. At high altitude temazepam reduces periodic breathing during sleep without an adverse effect on next-day reaction time, maintenance of wakefulness or cognition. The 2% reduction in mean SaO2 post-temazepam is likely to be predominantly because of acclimatization, as by chance more trekkers took temazepam on the first night (19 versus 14). We conclude that at high altitude temazepam is effective in reducing periodic breathing, and is safe to use, without any adverse effect upon next-day performance.


Sleep disruption at high altitude is almost universal (Anholm et al., 1992), because of a combination of unfamiliar, harsh living conditions, periodic breathing and hypoxia per se (Reite et al., 1975). Temazepam has previously been shown to reduce periodic breathing (Dubowitz, 1998) and to improve sleep efficiency and subjective sleep quality (Nicholson et al., 1988) at high altitude. Despite this, the mountaineering world has been reluctant to condone use of a benzodiazepine because of concerns regarding ventilatory depression at altitude, and indeed blunting of the ventilatory response to carbon dioxide following diazepam has been demonstrated (Utting and Pleuvry, 1975). Concerns regarding decrements in tasks requiring concentration and coordination have also been raised (Powles and Sutton, 1983).

Temazepam at altitude could theoretically improve next-day performance through reduction of periodic breathing – it has been shown at least at sea level that sleep fragmentation because of periodic breathing correlates with sleepiness (Bennett et al., 1998) and impaired performance (Bonnet, 1985). Alternatively it could impair next-day performance, as has been demonstrated in some benzodiazepines (Koelega, 1989).

Benzodiazepines are thought to reduce periodic breathing by decreasing arousals during sleep (Bonnet et al., 1990). At altitude, hypoxia stimulates ventilation, and PaCO2 may drop below the level required to stimulate ventilation, the so-called apnoeic threshold (Dempsey and Skatrud, 1986). Therefore, during sleep the relative importance of humoral drives to breathe increases, an apnoea may ensue with more marked hypoxia. Central apnoeas with this marked hypoxia are frequently followed by arousals (Sharp et al., 1986), with increases in heart rate, ventilatory rate and depth, increased oxygenation, and hypocapnia again, which help to sustain periodic breathing (Khoo et al., 1996). A brisker hypoxic ventilatory response is likely to produce a greater overshoot in ventilation, so leading the cycle to repeat itself (West et al., 1986), and indeed a correlation has been demonstrated between HVR and periodic breathing (Lahiri et al., 1983). Patients with central sleep apnoea have apnoeas two to three times more frequently in the transitional stage of sleep – stage 1 – than in stage 2 or rapid eye movement sleep (Bradley et al., 1986). Benzodiazepines have been shown to reduce stage 1 sleep, arousal frequency and sleep stage changes in both insomniacs (Dujardin et al., 1998) and patients with congestive cardiac failure (Biberdorf et al., 1993) and to increase stage 2 sleep in volunteers at altitude (Nicholson et al., 1988).

Using a double-blind, randomized cross-over trial of low-dose temazepam versus placebo at high altitude, we tested the hypotheses that temazepam is efficacious in reducing periodic breathing, and is safe, with no adverse affect on mean nocturnal SaO2, next-day performance or acute mountain sickness (AMS) score. The study setting was a field expedition undergoing a typical safe ascent profile for a Himalayan trek.


Trekkers underwent a 17-day graduated ascent in eastern Nepal from Tumlingtar (410 m) to a remote base camp near Chamlang (5000 m), on a high-altitude research expedition with ‘Medical Expeditions’ in April–May 2003. The ascent profile is illustrated in Fig. 1. They were invited to take part, and studied with written informed consent (Royal Brompton and Harefield NHS Trust). Exclusion criteria were a current acute illness, or high-altitude cerebral or pulmonary oedema, a known pre-existing sleep disorder and additional acclimatization because of an indirect ascent to base camp. Prior to the expedition trekkers carried out all tests (home nocturnal pulse-oximetry and actigraphy, reaction time, maintenance of wakefulness (MWT) and cognitive function testing) for familiarization purposes and to screen for sleep-disordered breathing.

Figure 1.

 Ascent profile of the trek from Tumlingtar (day 1: 410 m) to base camp (day 17: 5000 m).

Sample size

Our primary outcomes were reduction in periodic breathing following temazepam, and next-day performance (reaction time and maintenance of wakefulness). Assuming 90% power, 5% significance and a change in reaction time of 17 ms (as observed in healthy volunteers postsleep deprivation (Priest et al., 2001)), we calculated that the estimated sample size needed not to miss such a difference would be 15. We therefore aimed to study twice this number to allow for missing data.


A double-blind, randomized cross-over trial of 10 mg of temazepam versus placebo was carried out on two successive nights within 3 days of arrival at base camp. Each group of trekkers underwent block randomization to start with either temazepam or placebo, taken immediately before bedtime. The investigator carrying out randomization drew folded pieces of card out of a hat for each trekker and was not involved in subsequent collection of data or primary data analysis. Randomization was verified by an independent observer not directly involved in the project. Simple overnight sleep studies were carried out in the trekkers’ usual two-man sleeping tent. The next day the following were assessed: reaction time, MWT and cognitive function, subjective sleep quality and AMS score.

Timing of the daily routine was fairly constant during the trek because of communal living, and the constraints of daylight. Typical timings are shown in Table 1. Hours of sunlight restricted the timing of reaction time, MWT and cognitive function tests because of trekker comfort, operational temperatures of computers and other electrical equipment and availability of solar power for downloading data. Not all trekkers could be tested simultaneously for reaction time, MWT and cognitive function testing as only four sets of equipment were available; however, each trekker was tested at the same time of day following both temazepam and placebo.

Table 1.   Typical study timings
Time of day (hours)Activity
21:45Bedtime (in sleeping bag, ready to go to sleep)
06:45Get-up time
09:30–17:00Reaction time, MWT and cognitive function tests


Temazepam (10 mg) scored tablets were used (TEVA UK Ltd, Leeds, UK). Gut absorption is high at 90–100% , with peak plasma concentrations at approximately 50 min after ingestion, and an elimination half-life in the range of 7–11 h (mean 8 h) (Temazpam SPC Drug information).

Sleep studies

Trekkers wore a wrist oximeter with finger probe (Minolta, Pulsox-3i, Stowood Scientific Instruments Ltd, Oxford) on the non-dominant wrist, and movement sensor (Actiwatch®, Cambridge Neurotechnology Ltd, Cambridge, UK) on the other. At ‘bedtime’ trekkers turned on the oximeter and pressed an event marker on the actigraph; at ‘wake-up’ time trekkers turned off the oximeter and pressed a second event marker on the actigraph. SaO2 and heart rate were sampled every heart beat, with a 5-s moving average recorded. Actigraphy was recorded continuously, with data analysed in 1-min epochs. Automated software (Download 2001; Stowood Scientific Instruments Ltd, Oxford, UK) for SaO2 and heart rate analysis was used to determine mean SaO2, heart rate, drop in SaO2 of >4% below baseline (i.e. excluding those events equal to or less than 4%; (Pitson and Stradling, 1998) and heart rate rise index >6 bpm. Time spent in periodic breathing was determined manually from periodic desaturations; criteria were crescendo–decrescendo breathing lasting at least three cycles.

To determine the effect of Temazepam on SaO2 independent of periodic breathing, a separate analysis of SaO2 between two and four hours after bedtime, when temazepam would be expected to have its most noticeable effect upon sleep was carried out. Mean SaO2 during periodic breathing and non-periodic breathing was determined manually. Only data from trekkers with at least 10 min of suitable SaO2 trace (i.e. without excessive movements indicating wakefulness) were included.

Actigraphy, providing a minimally invasive method of distinguishing sleep from wakefulness (Sadeh et al., 1995), was used to determine sleep latency (time from bedtime to onset of a 10-min period without movement), actual sleep time (determined from an algorithm), assumed sleep time (time from estimated sleep onset to the wake-up time), sleep efficiency [(actual sleep time/time in bed) × 100), time in bed and the movement and fragmentation index, a marker of restlessness.

Reaction time and maintenance of wakefulness

Tests were carried out in a quiet, dimmed tent set away from the main camp. A sleeping bag, thermorest® and hot-water bottle helped maintain comfort, and headphones emitting white noise helped block extraneous sound. A 10-min multiple unprepared reaction time test was performed. Trekkers pressed a button as quickly as possible in response to a light emitting diode (LED) illuminating at pseudorandom intervals up to 10 s apart. The median reaction time was determined. The 40-min Oxford Sleep Resistance (OSLER) test (Bennett et al., 1998) was then performed. Trekkers lay comfortably looking up at the LED. They lifted a finger from a sensor each time it flashed on every 3 s. Sleep was concluded to have occurred if there was no response for 21 s, and the duration of prior wakefulness noted.

Cognitive function tests

A computerized testing battery (v2.2.0 CogState Ltd, Carlton, Vic., Australia, http://www.cogstate.com) was used, and carried out at sea level, and then again on the days following temazepam and placebo. Tests took 15–20 min to perform. CogState has been specifically designed for assessment of mild cognitive impairment, and is thought to be more sensitive in detecting mild cognitive deficits than conventional ‘pen and paper’ tests (Collie et al., 2001 (Makdissi et al., 2001), as might be expected with altitude and benzodiazepine exposure. Five variables were used for cognitive function testing as listed in Table 2, as they have suitable metric properties for serial study (i.e. they are normally distributed, without significant range limitation). Learning effects are stable after two repetitions of the test battery (Collie et al., 2001) and are minimized by a trial test prior to baseline testing and by the use of an infinite number of variable stimuli combinations (playing cards).

Table 2.   Five variables used for cognitive function testing
Cognitive function tests
Log mean of simple reaction time tests – psychomotor skills
Log mean of choice test times (‘red’ or ‘black’) – decision making
Total errors in ‘one back’ test (‘same’ or ‘different’ to last card) – working memory
Total errors in matching task – complex attention
Total errors in learning task – learning

Subjective measures

Acute mountain sickness was evaluated using the Lake Louise scoring system (Hackett and Olez, 1992). Following the sleep studies, trekkers indicated how quickly they dropped off to sleep compared with the previous few nights – quicker, the same or slower. After the second study they indicated their better night's sleep. Visual analogue scores (from 0 to 100 mm) were used to score sleep (‘worst night's sleep ever’ to ‘best night's sleep ever’) and tiredness (‘most tired ever’ to ‘no tiredness at all’).


Forty-one of 55 trekkers underwent randomization and entry into the study. Thirty-three had complete data sets, comprising 19 men and 14 women [median age 31 (range 23–72) years; body mass index of 22.1 (19.7–28.4) kg m2].

All trekkers were fit and healthy, although some had stable, treated conditions including hypertension (n = 1 on lisinopril 5 mg OD), type 1 diabetes mellitus (n = 1, on a continuous subcutaneous insulin infusion) and asthma (n = 2, on inhaled beclomethasone). Several were on an oral contraceptive pill. Five trekkers chose to take acetazolamide, 250 mg of BD during the ascent, as prophylaxis against altitude-related illness, although these individuals were not known to be particularly prone to altitude illness. One trekker had travellers’ diarrhoea and one had high-altitude pulmonary oedema a week prior to the start of the study, although both had recovered.

Ten of 33 trekkers were self-reported snorers, but none had a history suggestive of sleep-disordered breathing. On pulse-oximetry screening prior to the expedition at sea level, the median overnight SaO2 was 96.5% (range 92.2–97.7%), and the median 4% dip rate was low (1.6 per hour, range 0–5.3 per hour). The median heart rate rise index was 10.2 per hour (range 8.5–53.2 per hour).

By chance because of uneven data losses at altitude, slightly more trekkers took temazepam on the first night (19) than placebo (14). Reasons for data losses are shown in the flow diagram (Fig. 2).

Figure 2.

 Flow chart of trekkers participating in study (HAPE, high-altitude pulmonary oedema).

The median numbers of cups of caffeinated drinks taken by trekkers from wake-up time to performing reaction times and MWT at altitude was 2.0 following placebo and 1.0 following temazepam, with most trekkers taking the same number on both days.

Statistical analysis

Statistical analysis was carried out using Wilcoxon's signed-rank test to compare continuous data following placebo and temazepam, as several parameters were non-parametric (mean activity, sleep latency and 4% dip rate; Kolmogorov–Smirnov test). All results are presented as median (range) values. The effect of gender and age (less than or greater than the median age of 31.0) was examined using the Wilcoxon's signed-rank test.

A chi-squared test was used to analyse non-continuous data (proportions of trekkers failing to maintain wakefulness for 40-min and proportions reporting a better night's sleep). Subset analysis by sex (19 men and 14 women) and age (19 younger trekkers of less than or equal to the median age of 31 years, versus 14 older trekkers of greater than the median age) was carried out.

For statistical analysis of CogStateTM, a difference score from sea level (xSL) to base camp following placebo or temazepam (xP/T) was calculated for each subject in each of the five variables. The standard deviation (SDDiff) of this score for the placebo group was used as the denominator in calculating a z-score for each subject in each test:


A score of less than zero indicated worse performance following placebo or temazepam compared with sea level, and a score of greater than zero indicated better performance.

A value of P < 0.05 was considered significant for all tests, apart from CogState tests where multiple comparisons were made. For CogState, a value of P < 0.01 was said to be significant [0.05/(number of comparisons) = 0.05/5 = 0.01].


Efficacy of temazepam on periodic breathing and sleep quality

Overnight oximetry and heart rate in one trekker is illustrated in Fig. 3. Following placebo, periodic breathing persisted throughout most of the night. Following temazepam it was greatly reduced, only returning towards the end of the night; time-based analysis of 4% dip rate is illustrated in Fig. 4. After bedtime (0–2 h) the 4% dip rate was similar following placebo and temazepam; there was a progressive significant rise through the night following placebo (P = 0.002, Friedman test) and temazepam (P = 0.019). This occurred early in the night following placebo (0–2 compared with 2–4 h, P = 0.002, Wilcoxon's signed-rank test). In contrast, it occurred later in the night following temazepam (2–4 h compared with 4–6 h, P = 0.019). Temazepam was also associated with a significant decrease in the proportion of the night in periodic breathing, from 16.0 (0–81.3)% following placebo to 9.4 (0–79.6)% (P = 0.016, Table 3).

Figure 3.

 Nocturnal SaO2 (top line in red) and heart rate (HR, bottom line in blue) in one male trekker overnight (age 50 years), following placebo (top panel) and temazepam (bottom panel). Eight hours of each study is shown, with each line representing a successive hour of the night. For each hour (one line) the SaO2 and HR scales are the same, as indicated.

Figure 4.

 Median (IQR) 4% dip rate during hours 0–2, 2–4 and 4–6 of the night for placebo (dashed line; open circles) and temazepam (solid line; closed circles).

Table 3.   Measures of sleep
  1. Median (range) data for measures of sleep following placebo and temazepam; bpm = beats per minute. Wilcoxon's signed-rank test has been used throughout. P < 0.05

Sleep timings
 Sleep latency (min)14.0 (0–87)10.0 (0–63)0.15
 Actual sleep (hours – decimal)7.25 (4.77–8.77)7.05 (3.95–8.93)0.43
 Assumed sleep (hours – decimal)8.62 (5.65–10.1)7.92 (5.62–9.73)0.12
 Sleep efficiency (%)78.8 (58.3–91.4)76.9 (47.2–97.4)0.83
 Time in bed (hours – decimal)9.48 (6.85 10.78)9.50 (5.30–10.68)0.32
Pulse oximetry
 Mean SaO2 (%)78 (65–84)76 (64–83)0.01*
 Mean 4% dip rate (per hour)35.9 (5.5–113.3)31.1 (5.7–109.8)0.35
 Proportion time <80% (%)65.8 (2.3–99.7)80.1 (7.4–99.7)0.05
 Proportion of time in periodic breathing (%)16.0 (0–81.3)9.4 (0–79.6)0.02*
Heart rate
 Mean heart rate (bpm)66 (45–82)66 (51–82)0.36
 Heart rate rise index (>6 bpm per hour)42.6 (5.1–85.9)44.0 (13.0–82.2)0.11
Movement on actigraphy
 Mean activity (arbitrary units)9.1 (1.1–119.2)5.1 (1.3–47.9)0.33
 Movement and fragmentation index (arbitrary units)49.0 (10.4–104.3)44.0 (20.4–117.6)0.30

There were no significant differences in actigraphy-derived measures of sleep (sleep latency, actual sleep, assumed sleep, sleep efficiency or time in bed and fragmentation index, Table 3). Following temazepam compared with placebo, more trekkers reported dropping off to sleep quicker than usual (13 trekkers on temazepam versus nine trekkers on placebo), and sleeping better (17 trekkers on temazepam versus 14 trekkers on placebo). These differences were not significant.

Safety of temazepam on nocturnal SaO2, next-day performance and AMS

No one reported side-effects that were known to be attributable to temazepam. The reduction in periodic breathing following temazepam occurred at the expense of a small but statistically significant reduction in mean nocturnal SaO2 from 78 (65–84)% to 76 (64–83)% (P = 0.013, Table 3), with a borderline significant increase in the time with SaO2 <80% (P = 0.05, Table 3). Speculative reasons for the difference in SaO2 include the effect of periodic breathing itself, and the effect of acclimatization. However, less periodic breathing post-temazepam did not account for the difference in SaO2, as SaO2 during periodic breathing postplacebo was identical to that which occurred during non-periodic breathing, at 76% during hours 2–4 of the night (P = 0.12). During non-periodic breathing, as for the mean SaO2 for the whole night, SaO2 was significantly lower post-temazepam than placebo at 75% versus 76% (P = 0.013). Another possible reason for the difference in SaO2 is the effect of acclimatization, because by chance more trekkers took temazepam on the first night (n = 19) than on the second night (n = 14). Analysis was therefore carried out with equal numbers taking temazepam on the first and second nights, by omitting data from five trekkers studied post-temazepam on the first night and placebo on the second. The trekkers whose data were omitted were the first five trekkers to sign up for the expedition, and were from several different trekking groups. With 14 trekkers taking temazepam and placebo on the first night, there was still a significant reduction in the proportion on the night spent in periodic breathing post-temazepam [8.9 (range 0–63)% of the night compared with 17.1 (0–60.7)% of the night; P < 0.01]; however, there was no significant difference in mean SaO2 [77 (64–83)%, compared with 78 (65–84)%; P = 0.12]. The lack of a statistically significant difference in SaO2 post-temazepam is not likely to be simply because of the small reduction in sample size, as the number required to detect a difference in SaO2 of 1% is very large at 188, assuming 90% power and significance at the 0.05 level.

Twenty-one trekkers carried out reaction time and MWT testing in the morning and 12 in the afternoon (time range 09:21–17:35 hours). Reaction time was similar following temazepam and placebo, at 242 (204–386) and 241 (201–380) ms respectively (P = 0.84, Table 4). Fewer trekkers failed to maintain 40 min wakefulness following temazepam than placebo (four versus seven). This difference was not significant (P > 0.05, Table 4).

Table 4.   Next-day performance and symptoms
  1. Median (range) data for reaction time, maintenance of wakefulness, ‘z-scores’ for five domains of cognitive function during CogStateTM testing and acute mountain sickness (AMS) scores following placebo and temazepam. Wilcoxon's signed-rank test has been used, apart from where * is denoted, when chi-squared tests were used. A value of P < 0.05 was said to be significant for all parameters apart from cognitive function testing in which multiple comparisons were made, and P < 0.01 was said to be significant.

Next-day reaction time and MWT
 Reaction time (ms)241 (201–380)242 (204–386)0.84
 Maintenance of wakefulness (min)40 (8.5–40)40 (8.2–40)0.26
 Proportion failing 40 min MWT7/33 (21%)4/33 (9%)>0.05*
Next-day cognitive function testing
 Log mean of simple reaction time test0.07 (−3.6–1.33)0.06 (−2.07–2.00)0.72
 Log mean of choice test times0.19 (−1.29–2.24)−0.04 (−1.14–2.00)0.08
 Total errors in ‘one back’ test0 (−3.73–2.67)0 (−3.20–2.67)0.48
 Total errors in matching task0 (−5.50–3.93)0 (−3.93–2.36)0.21
 Total errors in learning task−0.07 (−1.86–3.14)0.07 (−2.57–3.43)0.54
Next-day subjective acute mountain sickness
 Better night's sleep?14/3317/33>0.05*
 AMS Total1.0 (0–1.0)1.0 (0–2.0)0.07
 AMS Tiredness0 (0–0)0 (0–1.0)0.058
 AMS Sleep0 (0–1.0)0 (0–1.0)0.18

Cognitive function data, available for 26 trekkers, is presented in Table 4. There were no differences in any of the domains of cognitive function.

Acute mountain sickness scores were surprisingly low (median = 1; range = 0–4, Table 4). However, there were small but consistently lower AMS scores in trekkers following temazepam compared with placebo, with a trend post-temazepam towards lower total AMS scores (P = 0.07) and tiredness score (P = 0.058) but not sleep disruption (P = 0.18). There was no significant difference in subjective quality of sleep using the visual analogue score (median score 28 mm, range 0–78 mm following temazepam, compared with 32, range 3–79 mm following placebo, P = 0.38), and no difference in tiredness (median score 14 mm, range 0–73 mm following temazepam, and 17 mm, range 0–82 mm following placebo, P = 0.23).

Effect of acclimatization

Median nocturnal SaO2 rose significantly from 76.0 (65–83)% on the first night to 78.0 (64–84)% on the second night (P < 0.001), independent of whether placebo or temazepam was taken. There were no changes in any other measures.

Effect of sex and age

Following placebo, men had a higher 4% dip rate than women (median 39.7, range 11.7–113.3 per hour compared with median 19.6, range 5.5–82.8 per hour, P = 0.024), and spent a greater proportion of the night in periodic breathing (median 20.6, range 0–81.3% compared with median 7.0, range 0–48.5%, P = 0.042). Older trekkers also had a trend towards a higher 4% dip rate than younger trekkers (median 53.2, range 5.8–113.3 per hour compared with median 31.2, range 5.5–82.8, P = 0.069), and spent a significantly greater proportion of the night in periodic breathing (median 25.1, range 0–81.3%, compared with 12.4, range 0–33.7%, P = 0.042). There were no significant differences in the effect of temazepam on men compared with women, or older compared with younger trekkers.

Effect of acetazolamide

The results obtained were unaltered by analysis with exclusion of the five trekkers who had taken prophylactic acetazolamide. Post-temazepam median SaO2 was still lower (76% versus 78%; P = 0.011), the proportion of the night with SaO2 < 80% was greater (83.7% versus 73.1%; P = 0.036), and the proportion in periodic breathing was less (9.9% versus 18.4%; P = 0.029).


This study is the largest to have examined the effect of a benzodiazepine on sleep at altitude, next-day MWT, reaction time and cognitive function, and include a significant proportion of women. Temazepam was effective in reducing periodic breathing. It was also safe, with no adverse effect upon next-day reaction time, maintenance of wakefulness, cognitive function or AMS. When analysis was carried out with equal numbers of trekkers taking temazepam on the first and second night to overcome the effect of acclimatization, there was no significant difference in mean SaO2 post-temazepam or placebo.

Limitations of this study

The aim of this study was to examine the efficacy and safety of temazepam during a ‘real-life’ Himalayan expedition, adopting a typical graduated, safe ascent profile, rather than a very acute ascent. This allowed partial acclimatization that may have diminished periodic breathing and so the effects of temazepam. The results will, however, be more applicable to future trekkers considering taking temazepam. The study was carried out in a relatively hostile, remote environment with large temperature swings over the course of the day, limited electrical power supply and short time frames available leading to a high intensity of studies (123 in 13 days). The study design was therefore simplified to be achievable in this environment. Overnight pulse oximetry and actigraphy was used rather than electro-encephalogram recordings, which we believe were sufficiently sensitive to address the aims of this study. At altitude all tests were carried out on only one occasion after each of temazepam and placebo because of time constraints placed upon the trekkers (many only stayed at base camp for 48–72 h, during which time they participated in other research projects), and availability of equipment.

The trekkers studied were a heterogeneous group, consisting of men and women, with a wide age range. The study was not powered to enable subset analysis; however, 4% dip rate and the proportion of time spent in periodic breathing was greater in men than women, and older compared with younger people, as has been observed to occur independent of anthropometrics in previous studies (Browne et al., 2001; Pack et al., 1992; Schneider et al., 1986). In our group, there was no correlation between 4% dip rate and either BMI (r2 = 0.04; P = 0.26) body weight (r2 = 0.02; P = 0.48). There were no significant differences in the effect of temazepam on any parameters measured either by sex or age, and therefore the results of this study are applicable to the general population of trekkers.

More trekkers were successfully studied using temazepam on the first night and placebo on the second than vice versa (19 versus 14) because of uneven data losses. As blood gases (decreased PCO2 and increased PO2), sleep quality, AMS and performance all improve with sojourn at altitude because of acclimatization, higher SaO2 and less periodic breathing would have been expected on the second night. Higher SaO2 was indeed seen (P < 0.001). On the second night, we would expect acclimatization to lead to less periodic breathing, but the greater use of placebo to lead to more periodic breathing. In actual fact, there was no difference in periodic breathing between the first and second nights (9.8% versus 12.4% of the night; P = 0.30). We think it likely that the higher SaO2 observed on the second night was predominantly because of the effect of acclimatization, as the difference disappeared when data were reanalysed with the same number of trekkers in each group.

Efficacy of temazepam on periodic breathing and sleep quality

Temazepam resulted in a modest but significant reduction in periodic breathing during sleep. This supports the preliminary findings reported by an author of this paper (GD) showing a reduction in periodic breathing of altitude with temazepam (Dubowitz, 1998). Our data also support the trends observed with zolpidem (Beaumont et al., 2004), and periodic breathing at sea level using triazolam (Bonnet et al., 1990).

The effect of temazepam wears off as the night progresses because of its short half-life of 5–8 h, so providing maximal benefit during sleep and minimal next-day residual effects. A low 4% dip rate was maintained for the first 4 h of the night following temazepam, before increasing at hours 4–6. Temazepam had no beneficial effect on sleep latency, as might be expected given the relatively short sleep latency even with placebo. Furthermore, the tablet form used in this study is more slowly absorbed than the soft gelatin form and has a longer half-life (Nicholson, 1989); it would therefore be expected to primarily benefit sleep maintenance rather than sleep onset, although this was not demonstrated in our study as actual sleep times were the same.

The stabilization of sleep stage mediated by temazepam may be the mechanism for the reduction in periodic breathing seen in the present study. If this were so we would have expected indicators of arousal (heart rate rise index and movement and fragmentation index) to be reduced during the first half of the night following temazepam. They were in fact unchanged. However, it is possible that the markers we derived from actigraphy were insufficiently sensitive to detect arousals, particularly as micro-arousals, as opposed to awakenings, can occur in the absence of movement.

Safety of temazepam on nocturnal SaO2, next-day performance and AMS

Following temazepam mean nocturnal SaO2 was statistically significantly lower than placebo by 2% . There are a number of possible causes for this. Firstly, following temazepam there is less periodic breathing. Salvaggio et al. (1998) showed SaO2 levels during periodic breathing to be greater than during regular breathing by 2.8 ± 1.7% , with no change in lowest SaO2 levels, and Lahiri et al. (1983) observed periodicity of SaO2 to be asymmetrical, with the crest being broader than the trough. This was not the explanation in our study, with identical SaO2 during periodic and non-periodic breathing postplacebo, and indeed periodic breathing only present for a small proportion of the night. A second explanation is that temazepam may have led to increased time spent asleep, which would decrease mean SaO2 as SaO2 is lower when asleep than awake. Actigraphy, however, did not indicate there was any difference in time asleep between temazepam and placebo. Thirdly, as more studies were carried out (by chance) post-temazepam on the first rather than the second night; reduced acclimatization (the ‘order effect’) may have contributed to the lower SaO2. This is likely to be the predominant mechanism, as the difference in SaO2 became non-significant when data from five trekkers were randomly omitted so that there were equal numbers taking temazepam on the first and second nights. Lastly, temazepam may have damped the hypoxic ventilatory response and reduced overall ventilatory drive so lowering mean SaO2 even in the absence of periodic breathing. In support of this last point, SaO2 was significantly lower post-temazepam than placebo between hours 2 and 4 of the night, even in the absence of periodic breathing (75% versus 76%; P = 0.013); however, this difference became non-significant when data from five trekkers were randomly omitted so that there were equal numbers taking temazepam on the first and second nights (P = 0.18).

In our study, older people (aged greater than the median age of 31.0 years) had more periodic breathing than younger people. However, older people would be expected to have a blunted HVR, which in turn should lead to less periodic breathing. This may have been because measurement of HVR performed at sea level neither showed difference between younger and older trekkers (P = 0.35), nor was there any significant linear regression with age (r2 = 0.02; P = 0.47). We therefore suggest that the effect of age on sleep structure is the dominant factor rather than the influence of age on HVR.

Whether a brisk HVR is ‘a good thing for a mountaineer’ is a debate that has continued for over 20 years (Milledge, 1986). On the one hand, Schoene et al. (1984) observed that climbers with a brisk HVR on Mount Everest attained higher altitudes, slept higher and had reduced exercise oxygen desaturation. On the other, Hornbein et al. (1989) on the same expedition observed that these mountaineers had a greater degree of cognitive impairment following exposure to extreme altitude, and speculated that this may be because of decreased cerebral blood flow because of hypocapnia. Additionally, Bernardi et al. (2006) have recently demonstrated that mountaineers who successfully summitted Everest or K2 without supplementary oxygen had less brisk HVRs, so giving them more ventilatory reserve between actual ventilation and maximum voluntary ventilation, and a more efficient breathing pattern with a lower dead space: alveolar space ventilation ratio.

Temazepam had no adverse effect on next-day reaction time, MWT or five measures of cognitive function, and indeed more people maintained wakefulness for 40-min following temazepam than placebo. We speculated that intermittent severe hypoxia during SaO2 dips during periodic breathing might have a detrimental effect on performance, even in the absence of a change in mean SaO2; however, there was no relationship between next-day performance and periodic breathing. The absence of a residual next-day effect may be because of the short half-life and low dose of 10 mg used. Similarly, a previous study has shown zolpidem or zaleplon to have no adverse effect upon cognitive performance or attention at simulated high altitude (Beaumont et al., 2004). A review of the effects of temazepam on vigilance in volunteers without known periodic breathing showed 10 mg to have no adverse effect, whereas 20–40 mg did (Koelega, 1989). Interestingly, a study using temazepam 15 mg in patients with periodic breathing because of congestive cardiac failure resulted in a significant improvement in objective sleepiness as assessed using multiple sleep latency tests (Biberdorf et al., 1993). A review of temazepam on computer-based psychomotor tests performed at low altitude showed it to have highly variable effects (Kunsman et al., 1992).

It is perhaps surprising that objective sleepiness was not greater as an index of arousal (movement and fragmentation index and heart rate rise index) was high, and it is known 10-min sleep periods are required for sleep to be restorative (Bonnet, 1986). It may be that during centrally mediated sleep apnoea these indices are not associated with sleep fragmentation in the same way as during obstructive sleep apnoea. In addition, increased sleep opportunity enabled poor sleep quality to be compensated for by increased quantity. Speculatively, there may be an opposing force promoting wakefulness, such as hypoxia. Furthermore, highly motivated individuals can maintain wakefulness despite fatigue.

It had been suggested by mountaineers that use of a sleeping tablet such as temazepam might increase AMS; however, it was interesting to observe that there was a trend toward a decrease in overall AMS score and tiredness. The incidence of AMS was unusually low in our group, and so data post-temazepam in a group with more AMS because of a faster or higher ascent profile would be of interest.

Clinical implications of this study

In our cohort of trekkers who ascended slowly to 5000 m and used a low dose of temazepam of 10 mg, it was found to have a beneficial effect upon periodic breathing, and to have no adverse effect upon reaction time, MWT, cognitive function or AMS. There was, however, a high degree of variability between trekkers, and there may be individual susceptibility to both beneficial and adverse effects. Further questions remain regarding the efficacy and safety of temazepam given in larger doses, its safety in the event of unexpected rapid responses and decision making being required in the middle of the night before the effects of the drug have worn off, or following more rapid ascent and to higher altitudes.


We are extremely grateful to Stowood Scientific Instruments, Oxford, UK, and Cambridge Neurotechnology, for their technical support, and to Mike Skinner and Tracey Hughes for their contribution to sea level data collection. All studies were carried out in association with the charitable group that promotes high altitude research and education, Medical Expeditions (http://www.medex.org.uk).


This work was funded by grants from the Oxfordshire health services research committee, the William Gibson Scholarship for Medical Women (Royal Society of Medicine, London) and the Wellcome Trust.