Modafinil for excessive daytime sleepiness

  • Protocol
  • Intervention

Authors


Abstract

This is the protocol for a review and there is no abstract. The objectives are as follows:

To assess the efficacy and safety of modafinil and its congeners (armodafinil), in producing and sustaining wakefulness, in people with EDS across different diseases.

Background

Description of the condition

Excessive daytime sleepiness (EDS) is characterised by an increase in fatigue during the day despite adequate rest at night. EDS affects approximately 20% of the US population (Pagel 2009); in those affected, EDS is a cause of significant morbidity and decreased health-related quality of life (Briones 1996; Vignatelli 2004; Wu 2012). Daytime sleepiness can be particularly hazardous for professional drivers (Masa 2000). It also impairs physicians' performance (Chen 2008). EDS is a manifestation of many different aetiologies including insufficient sleep, sleep-related breathing disorders, periodic limb movements, primary disorders of alertness such as narcolepsy and idiopathic hypersomnia, and general medical conditions including diabetes, dementia, parkinsonism, depression. Increased daytime sleepiness can also be a consequence of consuming other drugs including, sedatives, antihistaminics and neuroleptics (Guilleminault 2001; Harris 2012).

Over the years, scholars have developed various tools to measure sleepiness, though each tool suffers from its own limitations. These include subjective outcome measures such as the Stanford Sleepiness Scale and the Epworth Sleepiness Scale (ESS) (Hoddes 1972; Johns 1991), and the more objective tests such as the Multiple Sleep Latency Test (MSLT) and the Maintenance of Wakefulness Test (MWT) (Carskadon 1986; Doghramji 1997).

Treatment of EDS depends upon the underlying aetiology. For example, there is strong evidence that treatments such as continuous positive airway pressure can reduce fatigue caused by moderate and severe obstructive sleep apnoea (Giles 2006). Other treatments target the sleepiness itself. Classically, psychostimulants such as amphetamines and methylphenidate have been used to relieve symptoms of daytime sleepiness; however, these drugs have a high potential for abuse (Harris 2012).

Description of the intervention

Modafinil (2-[(diphenylmethyl)sulfinyl]acetamide) is a novel eugeroic drug that acts as a 'wakefulness-promoting agent'. Chemically, modafinil is a racemate comprising of 50:50 mixture of the R- and S-enantiomers of which the R-enantiomer (armodafinil) is the longer lived enantiomer. Armodafinil and adrafinil (prodrug of modafinil) are separately available commercial congeners. The usual dosage range for oral administration is 200 to 400 mg/day.The clinical pharmacokinetics is dose independent over a dose range of 200 mg/day to 800 mg/day (Wong 1999).The maximum plasma concentration is reached in two to four hours and the steady state concentration in two to four days after oral administration (Robertson 2003). The elimination half-life of modafinil is 10 to 15 hours, which is similar to its R-enantiomer, while the S-enantiomer has a much shorter half-life. The drug is metabolised mainly in the liver and the metabolites are excreted in urine (Wong 1999; Robertson 2003).

Modafinil is currently approved by the US Food and Drug Administration (FDA) for use in excessive sleepiness associated with narcolepsy, shift work disorder and obstructive sleep apnoea (Valentino 2007). However, the 2010 recommendations of the European Medicines Agency (EMA) advise its use only in the treatment of narcolepsy (EMA 2011).

Its efficacy has been investigated in a wide variety of conditions, including Parkinson's disease (Högl 2002; Ondo 2005; Van Vliet 2006), cocaine dependence (Dackis 2005), schizophrenia (Turner 2004b), Huntington's disease (Blackwell 2008), attention-deficit/hyperactivity disorder (Turner 2004a; Boellner 2006; Lindsay 2006), fatigue (Rammohan 2002; Zifko 2002; Fava 2005), spastic cerebral palsy (Hurst 2002), and myotonic dystrophy (MacDonald 2002; Orlikowski 2009).

Modafinil has been shown to decrease food intake and has been reported to cause weight loss ((Makris 2004; Henderson 2005; Perez 2008). Several studies also report that modafinil enhances cognition in healthy volunteers and in people with chronic schizophrenia (Turner 2003; Turner 2004b; Müller 2013), attention deficit hyperactivity disorder (Turner 2004a), and Huntington's disease (Blackwell 2008).

In comparison to alternative medications such as amphetamines, modafinil is generally considered as a safe drug with minimal or negligible adverse effects. However, study participants have discontinued its use due to serious adverse consequences such as headache, nausea, anxiety, dizziness, insomnia, chest pain, and nervousness. Other commonly reported harmful effects include, rhinitis, diarrhoea, back pain, anxiety, dizziness and dyspepsia. In people with psychiatric history, some have developed mania, delusions, hallucinations, and suicidal ideation. Post marketing surveillance has reported cases of multi-organ hypersensitivity reactions, life-threatening rashes and Stevens-Johnson Syndrome (EMA 2011; Modafinil 2013).

Modafinil is relatively affordable and is available in different brand names (such as Provigil® and Modalert®) from different manufacturers. The retail price ranges from about USD$2 to USD$10 per pill, depending upon the country and manufacturer (Everydayhealth 2011).

How the intervention might work

Though the exact mechanism of action remains unclear, modafinil is known to act in ways, and at sites, which are distinct from the classical 'stimulants' such as amphetamines (Lin 1996; Engber 1998). Modafinil has been shown to act through a variety neurotransmitters (dopamine, histamine, noradrenaline, gamma-aminobutyric acid (GABA), serotonin) at various locations especially in the hypothalamus (Tanganelli 1995; Ferraro 2000; Scammell 2000; De Saint 2001; Gallopin 2004; Dopheide 2007; Ishizuka 2010). In recent years, researchers have explored modafinil's neuroprotective actions. These properties might be relevant to its wakefulness-promoting activity (Xiao 2004; Gerrard 2007).

Modafinil may produce age-dependent effects, especially in relation to neurological outcomes, probably due to differences in baseline activity levels, metabolism and other age-related changes (McFadden 2010). Underlying causative conditions and point of follow-up also may cause differences in outcomes.

In comparison to other stimulants, modafinil appears to be free from most of the undesirable effects. It is of limited abuse potential as individuals are unlikely to become dependent on it. However, it induces hepatic enzymes and as a result it can interact with several other drugs including contraceptives, antiepileptic drugs, and antidepressants. Furthermore, it has the potential to harm to the foetus when taken during pregnancy (Valentino 2007).

Why it is important to do this review

Studies suggest that modafinil (1) has specific use in sleep disorders, (2) may potentially have several neurological benefits and (3) be of use in several neurological disorders (Valentino 2007; Repantis 2010). It is considered to be a safe drug, but concrete information regarding this is limited, especially with harms and long-term outcomes. A Cochrane systematic review evaluated the use of modafinil to relieve EDS in people with myotonic dystrophy (Annane 2009). The review concluded that there is low quality evidence available that suggests modafinil does not significantly improve results from MWT but that it does significantly improve the results from ESS. A non-Cochrane systematic review evaluated the role of modafinil in narcolepsy and concluded that it is effective in terms of reducing daytime sleepiness (Golicki 2010). We wish to undertake this review in order to estimate the effect of modafinil from randomised controlled trials, including quasi-randomised trials in people with EDS, across underlying disease conditions.

Objectives

To assess the efficacy and safety of modafinil and its congeners (armodafinil), in producing and sustaining wakefulness, in people with EDS across different diseases.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) and quasi-RCTs where the participants have been allocated using a quasi-random method (e.g. by date of birth, day of the week, month of the year, medical record number).

Types of participants

We shall include people with EDS or with conditions which lead to EDS (e.g. narcolepsy, sleep apnoea, periodic limb movement disorder).

We shall exclude trials where modafinil has been used for to treat daytime somnolence produced by other drugs (sedatives, antihistaminics, neuroleptics etc). As there are two other Cochrane systematic reviews evaluating the efficacy of modafinil in treating EDS caused by shift work sleep disorder and myotonic dystrophy (Annane 2009; Liira 2012), we shall exclude trials in these populations.

Types of interventions

Modafinil or congeners (armodafinil, adrafinil) at any dose or regimen compared with placebo, no intervention or other interventions (amphetamines) to maintain wakefulness.

Types of outcome measures

Primary outcomes
  1. Wakefulness/sleepiness (as reported in the trial)

    1. Episodes of falling asleep and/or sleep attacks

    2. ESS

    3. MWT

    4. MSLT

  2. Serious adverse events (death/requiring hospital admission/requiring additional intervention/causing significant impairment or discontinuation of treatment)

  3. Functional performance (number of injuries, number of errors at work)

We will present the primary outcomes in a 'Summary of findings' table. For the wakefulness and/or sleepiness outcome, we will use the two measures 'episodes of falling asleep' and 'EES' in the table as these are the most patient-oriented.

Secondary outcomes
  1. Adverse events (insomnia, headache, nausea, as reported in the study)

  2. Quality of life (as measured in the trial)

  3. Cognitive ability (as reported in the trial), including construction, concept formation and reasoning, executive function and motor performance, memory, orientation and attention, perception, verbal functioning and language skills.

We shall group outcomes based on the post intervention time point of follow up, into short term (one week to one month) and long term (more than one month) and report the mean duration of follow-up.

Search methods for identification of studies

Electronic searches

We will search the following electronic databases:

  1. The Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library)

  2. Ovid MEDLINE(R)

  3. Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations

  4. EMBASE

  5. PsycINFO

  6. CINAHL (Cumulative Index to Nursing and Allied Health Literature)

  7. Science Citation Index

  8. Social Sciences Citation Index

  9. Conference Proceedings Citation Index- Science

  10. Conference Proceedings Citation Index- Social Science & Humanities

  11. Database of Abstracts of Reviews of Effects (DARE, The Cochrane Library)

  12. Cochrane Database of Systematic Reviews (CDSR, The Cochrane Library)

  13. ClinicalTrials.gov (http://clinicaltrials.gov/)

  14. metaRegister of Controlled Trials (mRCT) (http://www.controlled-trials.com/mrct/)

  15. World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) search portal (http://apps.who.int/trialsearch/)

We shall search for studies irrespective of publication status or language. Searches will be based on the following Ovid MEDLINE search strategy, which uses the Cochrane highly sensitive search strategy to find randomised trials (Higgins 2011). We will adapt the search terms for other electronic sources as appropriate.

  1. exp sleep disorders/

  2. (hypersomnolenc$ or hyper-somnolenc$).tw.

  3. (hypersomnia$ or hyper-somnia$).tw.

  4. (somnolence or sleepiness).tw.

  5. (excessive$ adj3 sleep$).tw.

  6. (day$ adj3 sleep$).tw.

  7. narcolepsy.tw.

  8. shift work sleep disorder$.tw.

  9. sleep apn?ea$.tw.

  10. or/1-9

  11. Central nervous system stimulants/

  12. Benzhydryl Compounds/

  13. (psychostimulant$ or psycho-stimulan$).tw.

  14. (wakefulness adj2 promot$).tw.

  15. benzhydryl$.mp.

  16. modafinil$.mp.

  17. (adrafinil or armodafinil).mp.

  18. (Provigil or Modavigil or Nuvigil).tw.

  19. analeptic.tw.

  20. or/11-19

  21. randomised controlled trial.pt.

  22. controlled clinical trial.pt.

  23. randomi#ed.ab.

  24. placebo$.ab.

  25. drug therapy.fs.

  26. randomly.ab.

  27. trial.ab.

  28. groups.ab.

  29. or/21-28

  30. exp animals/ not humans.sh.

  31. 29 not 30

  32. 10 and 20 and 31

Searching other resources

We will search the reference lists of included studies and related reviews to find additional records not found by the electronic searches. We will attempt to contact experts to ensure we have not missed any studies that could be included.

Data collection and analysis

Selection of studies

Two review authors (MP and SF) will independently screen the titles and abstracts of records obtained through searching. They will classify the records into included, excluded and awaiting classification based on the inclusion criteria and resolve any disagreements by arbitration with the third author (MEM). We shall obtain the full text reports of the studies awaiting classification. If we are not able to make a judgement on the basis of a review of the full paper, we will classify as 'pending assessment' and seek additional information from the study authors. We shall identify multiple reports from the same study and link them to the primary study.

Data extraction and management

Two review authors (TB and SF) will independently extract data from the included trials using a data extraction form. We shall extract data under the following headings.

  • General information: title; authors; source; country; year of publication; setting.

  • Trial characteristics: design; method of randomisation; concealment of allocation; blinding of participants, personnel and outcome assessors, potential conflicts of interest.

  • Participants: inclusion and exclusion criteria; underlying disease; sample size calculation; baseline characteristics; losses to follow-up.

  • Intervention and comparison data.

  • Outcomes: as mentioned above (primary and secondary outcomes).

  • Additional data: adverse events and outcomes not described above.

The extracted data will be compared and disagreements will be resolved by consulting the trial report and another author (MEM). The third author will then enter all the information into the Cochrane Collaboration's statistical software, Review Manager 2013, and perform the analyses.

Assessment of risk of bias in included studies

We shall assess the risk of bias in included studies using the Cochrane Collaboration's tool for assessing risk of bias (Higgins 2011). We shall make the judgements of high, unclear and low risk of bias, based on the guidelines of the Cochrane Collaboration, in the following domains.

  • Sequence generation: we shall assess the study report for mention of the method of random sequence generation. We shall consider sequences generated using predictable or non-random methods (date of birth and/or admission, hospital record number, alternative sequence) as having high risk of bias. We shall consider use of a random number table or computer generated random sequence or Coin/dice tossing or shuffling of cards/envelopes or drawing of lots, to be of low risk of bias.

  • Allocation concealment: we will assess the study report for clear statements regarding the method of allocation concealment. We shall consider methods involving central allocation using telephone and/or web-based and/or pharmacy-controlled randomisation or Sequentially numbered and identical drug containers or opaque sealed envelopes, to be of low risk of bias .We shall assign high risk of bias when allocation was not concealed or when the method was inappropriate and the random sequence was possible to be accessed by the person performing the allocation.

  • Blinding: we shall assess studies for adequacy of methods used for blinding and whether participants, investigators or outcome assessors were blinded. We shall assign risk of bias judgements in terms of performance and detection bias for outcomes based on the probability that the method or the lack of blinding could have influenced the results. We shall assess risk of bias for subjective and objective outcomes separately. When outcomes are objective and blinding is unlikely to bias results we shall assign a low risk of bias.

  • Incomplete outcome data: we shall assess data for completeness based on the proportion of participants who entered the trial and availability of data at the time of follow-up and those who completed the trial. We shall assign low risk of bias when there is no missing data or when reasons (as reported in the trial) for loss to follow-up or the proportion is small such that it is unlikely to influence the results. We shall also consider the relative proportions between the loss to follow up in intervention and control group and appropriate sensitivity analysis while making these judgements.

  • Selective reporting: we shall try to assess the trials for selective outcome reporting by comparing reports of the same trial from different sources (trial protocol, multiple publications, direct correspondence). We shall compare methods and reported results to assess whether data is reported for the entire population and for all outcomes observed. We shall also look for important or non-significant results that are inadequately reported or if common or important outcome variables were missed by any of the trials.

  • Other sources of bias: we shall look for design-specific sources of bias such as: recruitment bias; baseline imbalance; loss of clusters; incorrect analysis; source of funding; ethics committee clearance; Informed consent procedures, and washout period.

These judgements will be assigned independently by the two authors extracting data (SF and TB). Any disagreements will be resolved by consulting the third author (MEM) and by referring to the trial report.

Measures of treatment effect

Continuous data

When outcome data are continuous, we shall use the mean difference (MD) and corresponding 95% confidence intervals (CIs) to estimate the treatment effect. If outcomes are measured on different scales we shall calculate the standardised mean difference (SMD) and the associated 95% CIs. We shall report the pooled SMDs and make interpretations by converting the pooled estimate back into one scale used most commonly in the included studies. We shall use change scores in preference to final scores, but if only final scores are reported, we shall present them in separate analyses.

Dichotomous data

When outcome data are dichotomous, we shall attempt to extract the number of people experiencing the event, the total number randomised and number available at follow-up in that group. If event data is not available in this form, we shall attempt to calculate it from the available data or contact the authors for it or present them in separate analyses. We will use relative risks (RRs) and 95% CIs as it is easy to make interpretations from this effect measure.

Rate data

When outcome data are reported as rates, we shall extract rate ratios (in per person years) and associated 95% CIs.

The primary outcome of wakefulness and/or sleepiness can be measured as rates in episodes of falling asleep and as continuous scales/scores in the ESS and MSLT and MWT. Adverse events and functional outcomes may be reported as dichotomous data or as rates. Quality of life and cognitive ability is commonly reported as scores on their respective scales.

Unit of analysis issues

Cross-over trials

When an included trial is of cross-over design, we shall attempt to incorporate the results in meta-analyses using paired analysis by assuming there is no carry over effect. If the trial did not report paired analysis, but sufficient information is available (individual patient data, paired t test, treatment specific summaries, McNemar test) to calculate the within subject correlation, then we shall extract the corresponding information and adjust for the within subject correlation (Elbourne 2002). If this is not possible, we shall extract information from the first phase of the trial and consider it as a parallel group design. If such information is not available in the report, we shall correspond with the authors to obtain it.

The minimum washout period for modafinil is about four days (as calculated from: 5 x half life of modafinil = 5 x 15hrs = 3.125 days). But, the minimum acceptable wash-out period for all cross-over trials in general is seven days. We shall explore the impact of washout periods less than seven days by means of sensitivity analysis.

Trials with multiple treatment groups

When included trials have multiple treatment groups, we shall ensure that a participant group is not represented more than once in the same analysis. We shall try to make pair-wise comparisons by combining the treatment arms in ways that they are most similar and clinically appropriate. If this is not possible, we shall make a decision based on which pairs are the most clinically relevant and most similar to the recommended use and extract the data only for those groups.

Cluster-randomised trials

We do not expect to find cluster-randomised trials. However, if we do find such trials, in cases where the effect estimates are not adjusted, we shall extract the intracluster correlation coefficient provided in the trial report, or contact the authors for the information and adjust the effect estimate for the design effect (Higgins 2011).

Dealing with missing data

We shall attempt to obtain any information that is missing or unclear by contacting the authors (outcome data, summary data and missing individual data). We shall calculate loss to follow-up as the proportion of the number analysed to the number randomised for each outcome. We shall report the information in the risk of bias assessment and discuss its impact on the conclusions of the review.

We shall perform per protocol analysis on dichotomous outcome data. When standard deviations (SDs) are missing from continuous outcome data, we will either calculate these from other available data such as standard errors (SEs), or will impute them, if possible. We shall impute SDs on the basis of SDs for the same outcome using the same scale, from within our review or from other similar studies, reviews or meta-analyses (Higgins 2011).

Assessment of heterogeneity

When meta-analysis is possible we shall assess for statistical heterogeneity by visually examine the forest plots for overlapping confidence intervals and use the Chi2 test for homogeneity and the I2 statistic for heterogeneity to estimate the contribution of true heterogeneity. The Chi2 statistic will be considered to represent significant heterogeneity at a level of 10% significance (P value < 0.1). When using I2 to assess heterogeneity, it's values are sensitive to several factors and we shall explore their effect by the use of subgroups. We will consider an I2 statistic value of less than 40% to represent mild heterogeneity that can be ignored; 30 to 60% as moderate heterogeneity and 50 to 90% to represent significant heterogeneity, both of which can be explored using subgroups. We shall consider an I2 value of 75% or greater as substantial heterogeneity, and when we are unable to explain it by subgroups, we shall report the results of each study without performing meta-analysis.

Assessment of reporting biases

We shall assess for reporting or publication biases by visually examining funnel plots for asymmetry (provided that there are at least 10 included trials in an analysis). If we are able to identify unpublished studies in our search, we shall examine the effect of publication bias by comparing their pooled results with that of the published studies (Higgins 2011). We shall also explore other possible causes of funnel plot asymmetry like heterogeneity, other risk of bias.

Data synthesis

We intend to pool results from studies using the following comparisons.

  • Modafinil versus placebo/no treatment.

  • Modafinil versus amphetamines.

  • Modafinil versus other interventions.

  • Or other comparisons where modafinil congeners were used instead of modafinil.

In each of the analyses the intervention may be used at any dose or regimen but may be subgrouped or analysed separately based on statistical heterogeneity.

We shall attempt to pool results only when there is no substantial statistical heterogeneity (I2 > 75%) that cannot be accounted for by subgroup analysis. We shall use the fixed-effect model of analysis except in cases where there is a moderate level of heterogeneity that cannot be accounted for by subgroups. In such cases we shall use the random-effects model.

If outcome data is contributed by studies of varying designs (individually randomised, cluster randomised), we shall perform meta-analysis using adjusted (for clustering) effect estimates and the generic inverse variance method. When outcome data is reported in a range of follow-up periods across trials, we shall group them into time periods (as specified in subgroup analysis) and perform a stratified meta-analysis.

If meta-analysis is not possible, we shall perform a descriptive, qualitative critical appraisal of the outcome information from the included trials.

Subgroup analysis and investigation of heterogeneity

We shall explore clinical heterogeneity in various analyses using the following subgroups.

  1. Age.

  2. Underlying condition (e.g. shift work sleepiness, narcolepsy, sleep apnoea).

  3. Point of follow-up (to detect tolerance).

  4. Differences in intervention or controls in terms of dosage, regimen and placebo or no treatment.

Sensitivity analysis

We shall assess the robustness of assumptions and decisions made during the review process by performing sensitivity analysis with regard to loss to follow-up and for studies with missing outcome data. We shall examine the change in effect estimates by assuming worse-case and best-case outcomes among those lost to follow-up. We shall examine the effect of data imputation by removing such studies from the meta-analysis and observing the change in the pooled estimate.

We shall also perform sensitivity analysis to examine the differences in effect estimates caused as a result of different study designs or risk of bias.

Acknowledgements

We acknowledge the contributions of Dr. Prathap Tharyan (Director, South Asian Cochrane Centre) and Richard Kirubakaran (statistician, South Asian Cochrane Centre) for their invaluable help in drafting the protocol.

Contributions of authors

Preparing the protocol draft - MEM, TB
Verifying and approval of the protocol - SF, MP

Declarations of interest

Manu Easow Mathew - during the time of writing this protocol, was employed at the South Asian Cochrane Network and Centre (SACN) authoring systematic reviews, training and mentoring authors. He is no longer employed by SACN.

Tamoghna Biswas - was a student ambassador for Elsevier India in 2011 and is currently a country representative of Healthcare Information for All 2015, for India. Both are nonpaid positions.

Shabin Fahad - during the time of writing this protocol, was a tutor in the Department of Forensic Medicine, Kasturba Medical College, India.

Mitali Patnaik - none known.

Sources of support

Internal sources

  • South Asian Cochrane Centre, India.

External sources

  • No sources of support supplied

Ancillary