Moreover, the symptom of cataplexy seemed to go hand in hand with rapid REM onset.15 Several symptoms of narcolepsy, such as cataplexy, sleep paralysis, and hypnogogic hallucinations, could be interpreted as some aspect of REM sleep—dreaming or loss of muscle tone—intruding into waking time. The central problem was seen as a lack of properly regulated REM sleep. This view was adopted as the definition of narcolepsy at the First International Symposium on Narcolepsy in France in 1975: “A syndrome of unknown origin that is characterized by abnormal sleep tendencies, including excessive daytime sleepiness and often disturbed nocturnal sleep, and pathological manifestations of REM sleep.”

  At the time, little information was available about the neurological basis of narcolepsy. While investigators such as Dement were beginning to probe brain chemicals in both people and dogs with narcolepsy, REM sleep was the most tangible phenomenon that they had to grasp. The emphasis on REM onset shifted diagnosis away from relying on what patients recalled and toward examination in the sleep laboratory. Sleep specialists could use early REM onset as a way to confirm patients’ anecdotal reports of narcolepsy symptoms, such as sleep paralysis and hypnogogic hallucinations.

  THE GOLD STANDARD

  In the 1970s, Dement’s group at Stanford was having people with narcolepsy take single daytime naps to look for sleep-onset REM.16 However, the multiple-nap structure of the MSLT came from their work on healthy adolescents.17 His graduate student Mary Carskadon was experimenting with having volunteers go through a ninety-minute day—sixty minutes for wake and thirty minutes for sleep—in an attempt to outrun REM sleep. Dement and Carskadon found that REM appeared more quickly when people were deprived of it. They also observed that their subjects’ underlying sleepiness, driven by their internal circadian rhythms, varied throughout the day.

  In a phone interview, Carskadon said that she and Dement always had narcolepsy in mind while developing the MSLT, even if her work focused on healthy children and adolescents. She said they were influenced by the work of Bedřich Roth in Prague, who had also reported clinical experiments with single naps and REM onset. “Our goal was always to have a test for the clinic,” she said. “Once you have a tool that can measure physiological sleepiness, people can use it in many ways.”

  The idea behind the MSLT was straightforward. If someone fell asleep faster, that meant they were closer to the sleep state—and thus sleepier—beforehand. The person being tested was monitored while lying down in the dark, eyes closed, with no external obstacle to sleep put in their way.

  The MSLT’s developers presented it as objective,18 compared with simply asking someone how drowsy they are or measuring performance on an alertness task, which can be influenced by muscle fatigue, practice, or motivation. Someone can fill out a questionnaire however they like, but without the influence of sedatives or prior sleep deprivation, it’s more difficult to try to fall asleep faster than usual. Other sleep researchers, such as Tom Roth and his colleagues at Henry Ford Hospital in Michigan, were taking similar approaches to study the effects of antihistamines around the same time.19

  There were existing alternatives to measuring sleep latency, such as pupillometry—tracking the size of the pupil or its spontaneous oscillations, which reflect underlying signals in the nervous system. The Mayo Clinic had tested pupillometry for the diagnosis of narcolepsy beginning in the 1960s,20 but the approach required expensive equipment and seemed to be less reliable than the MSLT.21 Similarly, Canadian sleep researchers tested evoked potentials, a neuroelectrical technique sometimes used for diagnosis of multiple sclerosis. But neither approach looked for REM sleep, whose rapid onset was considered an important marker of narcolepsy.

  With sleep-onset REM in mind, Dement’s group at Stanford adapted the “experimental MSLT”—used for studying healthy young people—into the “clinical MSLT” for narcolepsy diagnosis. Four or five naps gave several chances for REM sleep to appear, within a cutoff of fifteen minutes. Two sleep-onset REM periods, plus an average interval before falling asleep of five minutes or less, were deemed sufficient. The five-minute average was later extended to eight minutes.22

  In the 1980s, the growing field of sleep medicine adopted the MSLT as a standard procedure for the diagnosis of narcolepsy. Although narcolepsy and cataplexy historically had been tied together, Dement and colleagues wrote: “There are unresolved differences of opinion as to whether a definitive history of cataplexy is a necessary component of the narcolepsy syndrome.”23

  A criticism of the MSLT emerged, reflecting the test’s operational basis: subjective sleepiness and the ability to quickly fall asleep do not necessarily match up. In some studies, various treatments for narcolepsy were effective in alleviating subjective sleepiness but not in extending sleep latency. In 1992, sleep researchers from the Netherlands referred to the MSLT as a “paradoxical test”: “This leads us to a situation where the very drugs, which are considered as the most potent in the treatment of sleepiness in narcolepsy, do not in fact result in an improvement of what is considered the most reliable test for sleepiness.”24

  Sleepiness has multiple dimensions, many researchers recognized. To extend the technique, Tom Roth and his group developed a variant of the MSLT called the Maintenance of Wakefulness test.25 A principal difference between the MWT and the MSLT is effort. In the MWT, participants are asked to sit up in a dark room and to try to stay awake, recruiting the parts of the brain that help someone deliberately resist sleep. The MWT has been used for medication studies and also for airline pilots or commercial drivers who need to show that they can stay awake.

  Divergent results for objective versus subjective sleepiness highlight an underlying issue. The MSLT measures one part of sleepiness, but it does not capture other aspects, such as those experienced by people with IH: the number of hours someone may need to sleep or the inability to leave sleep behind.

  NARCOLEPSY OUT ON THE STREET

  As a diagnostic test, what bothered Rye about the MSLT was that it often placed people into categories that didn’t make sense if one had more information about their medical conditions. “At the very least, the MSLT is a nonspecific test,” he said. “If you start looking at sleepy people besides those with classic narcolepsy, it doesn’t help you distinguish them.”

  Rye and Bliwise were the first to publish papers on sleepiness in people with Parkinson’s disease, showing that about 30 percent met MSLT criteria for narcolepsy.26 With Parker, they also established that a similar fraction of end-stage kidney disease patients also had pathological levels of sleepiness.27 Their 1997 survey of narcolepsy patients had used the MSLT, but it documented how people who didn’t fit the classic picture of the disorder were showing up in their clinic. It concluded: “Increased recognition of N- [narcolepsy without cataplexy] might reflect heightened public awareness to EDS [excessive daytime sleepiness] and its treatment options along with improved diagnostic capabilities.”

  In 1998, Rye and Bliwise published a case report on a forty-six-year-old woman with an apparent mixture of depression and narcolepsy.28 She experienced typical symptoms of depression and was prescribed antidepressants. However, she also had gained thirty pounds in the last several months and had displayed sleep attacks at work and while commuting. The woman didn’t report cataplexy, sleep paralysis, or hallucinations, but she did enter REM sleep in three out of five MSLT naps. Treatment with bupropion, an atypical wake-promoting antidepressant, alleviated both sleepiness and depression symptoms. Should the woman be seen as primarily having depression or narcolepsy? The boundaries were not clear.

  Rye wasn’t the first to point out the MSLT’s elasticity. In the early 1990s, other sleep researchers proposed that people whose clinical history didn’t fit narcolepsy should be labeled “hypersomnia with sleep-onset-REM periods” instead.29 Neither the distinction nor the clunky language stuck. In the United States, the FDA’s 1998 approval of modafinil for narcolepsy—without cataplexy as a necessary requirement—gave physicians and patients an incentive to move toward the narcolepsy diagnosis.

  The initial studies establishing the MSLT did not look at patients with sleep disorders besides narcolepsy. They also left out people who might be sleepy as a result of their work schedules. From their ninety-minute-day experiments, the Stanford researchers were aware that people without narcolepsy could display sleep-onset REM periods if deprived of REM in the laboratory. They may have discounted how often doctors would see the same effect “out on the street” (Rye’s phrase). It took years to organize studies that would include enough people to map that landscape, and the results were confounding.

  A community-based study in Wisconsin examined a group of more than five hundred people, not only those who came to a doctor’s office concerned about sleepiness. More than 5 percent of men and 1 percent of women met MSLT criteria for narcolepsy.30 This didn’t mean that there was an unseen epidemic of narcolepsy in Wisconsin but rather that the test roped in false positives. In particular, those with sleep apnea or who worked night-shift jobs showed up as apparently having narcolepsy. Tom Roth conducted a similar population-based study in Michigan, and he concluded: “If someone has a pathological level of sleepiness, you’re going to see sleep-onset REM some of the time.”31

  Researchers at the University of Pittsburgh noticed that people with depression displayed more rapid entry into REM sleep, but not as fast as in narcolepsy.32 Did that mean that depression had something in common with narcolepsy? Perhaps not. The significance of fast REM onset in people who don’t have narcolepsy remains unclear. The biological function of REM sleep is mysterious; people under the influence of certain antidepressants or with rare brain injuries can manage without experiencing much REM sleep at all.33

  In the 1990s and 2000s, sleep specialist clinicians used the MSLT with “unbridled enthusiasm,” according to an editorial in the trade publication Sleep Review: “In many sleep laboratories, the MSLT became part of the evaluation of virtually all sleepy patients, including those with obstructive sleep apnea, periodic limb movements, insomnia, and circadian rhythm sleep disorders, as well as patients with suspected narcolepsy and idiopathic hypersomnia.”34

  In 2005, a standards of practice committee of the American Academy of Sleep Medicine declared the MSLT “the de facto standard” for objective measurement of sleepiness yet warned that its diagnostic value, beyond confirming suspected narcolepsy, was limited.35

  THE MSLT EXPERIENCE

  The MSLT is something people with IH have in common; it is the gateway through which they had to travel to receive their diagnosis (figure 4.1). But listening to discussions at support group meetings or even lurking on social media, it becomes clear that sleep labs don’t always implement the MSLT according to current guidelines. Sometimes the patient does not sleep enough beforehand, because they are awakened early by external noise, incoming staff, or changing shifts. This may be a source of overdiagnosis, although a standard overnight sleep test lasts just seven hours, and someone with IH may normally sleep for much longer than that.36 On the other side of the coin, it is possible for someone who feels extremely sleepy to have trouble falling asleep on command. Performance anxiety seems to play a role.

  Normally, the night before an MSLT, an overnight sleep test checks for sleep apnea and leg movements and also measures how fragmented someone’s sleep is. An overnight sleep test collects a vast amount of information on someone’s EEG patterns and sleep stages, but that information often goes unexamined for narcolepsy/IH diagnosis.

  FIGURE 4.1. Both the PSG (polysomnogram) and MSLT (multiple sleep latency test) procedures can be uncomfortable.

  Source: Off the Mark Cartoons, Atlantic Feature Syndicate.

  Chronic insufficient sleep can make it look like someone has narcolepsy or IH, which is why recent guidelines call for actigraphy—wearing a device that monitors movement—for two weeks before an MSLT.37 Guidelines also call for patients to stop taking stimulants or antidepressant medications two weeks beforehand. Antidepressants can distort MSLT results because they suppress REM sleep, but physicians may be reluctant to insist on having someone discontinue antidepressants because of the possibility of withdrawal or worsening depression. Opioid or cannabis use can also confound MSLT findings. “The MSLT was one of the most stressful tests I could go through,” said David Kellogg, a licensed hearing aid specialist from Oregon who was diagnosed with IH in 2012.

  Several years before his IH diagnosis, David noticed that he was having trouble staying awake while driving to work. An overnight sleep study detected mild sleep apnea, so he underwent surgery recommended by an ear, nose, and throat specialist. It didn’t help. He would sleep ten to twelve hours per day and “still woke up feeling like a truck had driven over him.” He tried everything he could think of to feel better: exercise, eating better, vitamin D. He changed jobs, stepping down from demanding managerial roles. Sleepiness still threatened to pull him under while watching TV, reading, or even sitting at a red light.

  Unable to work, David went on disability, and he had to complete a second MSLT in 2016 to demonstrate that his IH was still present. Beforehand, he stopped both stimulants and antidepressants and experienced withdrawal side effects. In between naps, he was trying to relax but also texting his wife in alarm, worried that his disability payments were on the line. He still fell asleep five times, averaging less than four minutes.

  Another person who was diagnosed with IH in a Canadian hospital, and then later in the United Kingdom with narcolepsy, wrote: “The test was torturous: being woken repeatedly from the required naps left me with a violent migraine and I vomited into the clinic toilet.”38

  THE BUSINESS OF BREATHING AT NIGHT

  Another gateway some people with IH pass through is diagnosis with obstructive sleep apnea: fleshy, flabby interruptions in breathing that interfere with rest and put strain on the heart. High rates of obesity have made obstructive sleep apnea common in the United States, and the condition is the main revenue source for most sleep medicine practices.

  A standard remedy for obstructive sleep apnea is continuous positive air pressure (CPAP): having air blown up the nose to keep the airway open. Someone entering a sleep clinic is likely to see face masks and hoses on display. The first CPAP machines were driven by vacuum cleaner motors, but current devices have become more sophisticated, with internet connections and pressure modulation. Supplying devices for sleep apnea treatment is a multi-billion-dollar business.

  This leads us to one of Rye’s grumbles about his field. Many sleep specialists in the United States were trained as pulmonologists, not neurologists. According to Rye, they tend to manage sleepy patients with a “hammering everything that looks like a nail” approach, assuming obstructive sleep apnea is the cause.

  While CPAP provides benefits to many people, it can be uncomfortable; it has been estimated that more than half of those who are prescribed CPAP eventually abandon it.39 Complicating matters, untreated sleep apnea is thought to damage brain circuitry, contributing to excessive daytime sleepiness.40

  We might discount Rye’s complaints as arising from friction with colleagues or impatience with issues such as finding a face mask that delivers CPAP properly. Still, his counterparts concede that “pulmonary medicine specialists are often responsible for the diagnosis and treatment of a number of sleep conditions, including several that are not traditionally considered related to respiratory medicine.”41

  Others besides Rye have made similar criticisms. A sleep medicine program director in Maryland wrote in Sleep Review in 2014: “For two decades, clinical sleep medicine was somewhat myopically focused on diagnosing OSA, too often with little regard for the patient’s long-term adherence, satisfaction, or outcomes.”42

  David Kellogg’s experience in Oregon demonstrates how an assumption that a drowsy patient has sleep apnea can play out. David’s hourly rate of breathing interruptions was relatively low, but before his IH diagnosis, he still put in a solid effort at making CPAP work, trying different masks and pressures for eight months. At each appointment, he brought in the CPAP memory card to verify that he was using it.

  Even after his IH diagnosis, he faced doubts from some health care providers. At the advice of a council of academic physicians in Oregon, David had a second airway surgery that he was reluctant to undergo. He worried that refusing surgery would trigger a loss of disability payments. Afterward, his frequency of sinus infections went down, but he was still constantly tired. He said he felt like a car whose idle had been set too low and whose engine kept stalling. Only after his second MSLT did his new doctors agree that he had IH. “Nobody knew what IH was,” he said.

  Sleep specialists’ reluctance to go beyond an initial diagnosis of sleep apnea is understandable. They may not want to advance to prescribing stimulants, a possible consequence of an IH diagnosis, if someone hasn’t made a sustained effort with CPAP. Some recall patients who claimed to have a sleep disorder but seemed suspiciously interested in obtaining a prescription for stimulants.43

  Doctors have a saying: “When you hear hoofbeats, think of horses, not zebras.” That is, look for the expected cause, not something exotic. On average, many more people coming to a sleep clinic will have sleep apnea than narcolepsy or IH. Current estimates say that more than a quarter of men and 10 percent of women in the United States have some level of detectable disruptions of breathing during sleep.44

  Because it is so common, obstructive sleep apnea has a “borderland” issue akin to that of narcolepsy. With sensitive equipment that measures nasal air pressure, sleep labs have become very good at detecting disruptions of breathing. A 2015 study of more than two thousand middle-aged people in Switzerland found that three-quarters of the men and half of the women had obstructive sleep apnea, under International Classification of Sleep Disorders criteria.45 Higher than previous estimates, this was an “unrealistically high prevalence” and suggested a need to revise those criteria, the authors concluded.