Sleep

Sleep affects more than how rested someone feels the next morning. Small controlled studies have found that restricted sleep can reduce insulin sensitivity and glucose tolerance, showing a direct role in how the body handles blood sugar (Donga et al., 2010; Spiegel et al., 1999). These studies help explain why sleep duration matters for health.

The most common answer here is six to seven hours, chosen by 38.6% of responses. That band sits at or below the lower end of the seven-to-nine-hour range recommended for working-age adults; it does not separate six-hour nights from seven-hour nights. A further 30.7% select one of the two shorter bands. The picture therefore deserves attention beyond the very shortest nights, with sleep quality considered alongside duration. The recommendation is a guide, not a diagnosis (Hirshkowitz et al., Sleep Health, 2015).

The most common night is six to seven hours#

Of the 9,097 submitted survey responses, 7,569 answered how long they usually sleep. They selected one of the five ranges shown below.

Throughout this page, short sleep means one of the two shortest answers — under five hours, or five to six. That is the report's definition of the term, and it describes a band the respondent picked rather than a duration anyone measured. Someone who sleeps almost exactly six hours could reasonably choose either "five to six" or "six to seven".

2,327 answers — 30.7% — fell in the two shortest bands. Another 2,323, also 30.7%, selected seven-to-eight hours or eight-plus hours, the two bands that start at or above the lower bound of the seven-to-nine-hour range the US National Sleep Foundation recommends for working-age adults (Hirshkowitz et al., Sleep Health, 2015).

Neither is the biggest group. 38.6% chose six to seven hours, a band that reaches the lower end of the recommended range.

These answers place much of the sample at the shorter end of the duration choices. They do not establish the exact share sleeping less than seven hours, because the six-to-seven band includes its seven-hour boundary.

The sample is mainly salaried and skews towards IT and men. Methodology.

Shorter sleep bands account for more answers than in WHI 2025#

Short sleep is higher than either estimate available from WHI 2025. The rise is about 3.4 points against its published duration total, or 1.8 points against the estimate from its sex table.

WHI 2025 published a "less than 6 hours" category. WHI 2026 offers "under five hours" and "five to six" as its two shortest answers. These are an approximate comparison of the published shortest sleep bands, not a verified identical measure: the 2026 bands do not separate exactly six hours, and the 2025 boundary rule is not documented.

Against the overall duration list published in WHI 2025, the share selecting seven hours or more is 6.3 points lower and the six-to-seven-hour band is 2.9 points higher. The seven-plus difference applies to that published series only; other 2025 figures conflict with its upper bands.

The published bands point toward shorter reported sleep in the later sample, but the categories differ.

WHI 2025's overall and sex-specific tables do not reconcile. The 1.8–3.4-point difference uses those competing baselines; it is not a confidence interval. The seven-plus-hours baseline is also unresolved. Sex comparisons use the sex table's own figures.

One four-hour night cut insulin sensitivity by about a quarter#

Short sleep is not only a comfort problem, and the effects do not wait for years to accumulate. They are measurable within a day.

Both are small controlled experiments in healthy volunteers. They show what short sleep did to those volunteers. Longer-term risk has been studied in much larger observational samples: across 25 cohorts from 15 prospective studies and 1,381,324 participants, short sleep at baseline was associated with a 12% higher risk of death from any cause — a relative risk of 1.12, with a confidence interval running from 1.06 to 1.18 (Cappuccio et al., Sleep, 2010). Those cohorts sit inside a wider meta-analysis of 16 studies and 27 cohorts, covering 1,382,999 people followed for between four and 25 years.

None of this makes a six-hour night a diagnosis. The seven-to-nine-hour range is an expert consensus recommendation, not a clinical threshold, and nobody in the 38.6% has a condition by virtue of sitting there. Sleep duration can still matter below the point at which anything would be diagnosed.

Short sleep is flat through the 20s and 30s, then climbs after 40#

The first two decades of working life are within a point of each other, and the 30s are marginally the better of the two. The step comes after 40, and the 50s carry the highest short-sleep share of the four age bands.

The 50s band is also the smallest, at 229 answers. The four bands cover 7,509 of the 7,569 sleep answers; no band above 59 reaches the 50-answer minimum.

The shape is at least consistent with what objective measurement finds. Across 65 studies covering 3,577 healthy people from childhood to old age, adults get progressively less slow-wave sleep and spend more time awake after falling asleep as they age (Ohayon et al., Sleep, 2004). That is measured sleep in volunteers rather than an explanation of these answers, but it means an older cohort reporting shorter nights is not a surprise. WHI 2025 also found short sleep highest in its oldest band, though it used generational rather than decade boundaries, so the two editions cannot be compared band for band.

Women report short sleep eight points more often than men#

Eight percentage points is a difference in how often short sleep is reported. The two groups together account for 7,541 of the 7,569 sleep answers; 28 fall outside them.

Both editions found the same direction. WHI 2025's gender table — 1,125 women and 2,306 men — put the gap at 3.4 points, 31.2% against 27.8%. It is now 8.0 points. Almost all of the widening sits on one side: between the two gender tables the female share rose 5.1 points, while the male share moved 0.5.

A female excess in sleep complaints is a consistent finding in sleep epidemiology. A meta-analysis of 29 studies covering 1,265,015 people put the female-to-male risk ratio for insomnia at 1.41 (Zhang & Wing, Sleep, 2006). That research measures insomnia rather than these duration bands, and it does not explain the size of the gap here.

Why the gap is this large, this survey cannot say. It asked how long people sleep, not what fills the hours on either side of it.

Mumbai reports the most short sleep of the five largest metros#

Just under ten points separate the top of the table from the bottom: 9.8, Mumbai to Pune. These five cities carry 5,476 of the 7,569 sleep answers; the analysis publishes nine smaller city groups.

Mumbai's city page sets out the rest of its profile.

City rankings describe the people who answered, not the cities' workforces. No year-on-year city comparison is made here.

EdTech tops the short-sleep table; IT/SaaS sits 11 points below it#

Sectors with at least 100 answers, ranked. The residual "Other" bucket is not a sector and is not ranked.

11.2 points separate EdTech at the top from IT/Software/SaaS near the bottom, and it is the position of IT that is worth pausing on. It reports 27.6% on much the largest base of any sector — 2,807 answers, over a third of the total. Only Hospitality/Travel, on 130 answers, ranks lower.

Why IT sits where it does is not something this survey answers: there is no sector-by-age or sector-by-work-mode cut behind these rows.

The ranking is a 2026 reading rather than a property of these sectors. WHI 2025 profiled five sectors, on samples of 111 to 1,173. Manufacturing was the lowest of them at 24.1% and sits mid-table now at 30.4%. IT/Software was 28.9%, Healthcare 35.2%, BFSI 31.8%.

Pharma, Energy / Utilities and Government / Public Sector are not ranked because each has fewer than 100 answers.

One in four reported disturbed sleep in the last month#

How long someone sleeps is one question. Whether the night works is a different one, answered by 7,561 responses. Respondents could select more than one problem, so these shares overlap and cannot be added together.

These are the survey's own answer labels: "disturbed sleep in the last month" is the phrase the respondent selected, not a report of waking at any particular hour. The survey did not ask how often, for how long, or with what consequence the next day — the detail a clinician would need before calling any of this insomnia.

The four items did not move together. Disturbed sleep rose and now leads the list. Difficulty falling asleep fell by about four to five points and led the 2025 list under either estimate. Grogginess fell 3.3 points, and difficulty waking up moved 0.2.

Duration and quality can disagree. A quarter of these answers report a problem with the quality of the night, and a count of hours will never show it. One habit does track with the hours: short sleep runs from 25.1% of the responses reporting no caffeinated drink on a workday to 41.9% of those reporting four or more, which the Caffeine page sets out in full.

WHI 2025's overall and sex-specific sleep-quality tables give conflicting estimates. Both are shown where available, so the size of the edition difference is uncertain.

3.6% of sleep-aid answers reported alcohol to fall asleep#

A third question asked what people had used to help them sleep in the last month. It has a much smaller base than the others — 3,304 responses, under half the number who answered the duration question — so these are shares of that answer group and not of the workforce. Respondents could name more than one, so these shares overlap and cannot be added together.

Ashwagandha and magnesium head the list, at 5.5% and 4.8%, and both are reported more often than melatonin. The one prescription option is reported less often than either, at 2.8%.

118 responses reported reaching for a drink to get to sleep. It is a small share carrying a large clinical implication: alcohol works against the thing it is taken for. The Alcohol page resolves the same behaviour by drinking frequency: among weekly drinkers who answered both questions, it runs to 22.3%.

WHI 2025 published its own sleep-aid figures against its whole sample of 3,437 rather than an answer group, so no like-for-like comparison with this table is available in either direction.

Short sleep is more than twice as common in the high-stress responses as in the low#

The survey asks how long people sleep. Separately, it asks how stressed the respondent felt in the previous week, on a scale of one to ten. 7,343 responses answered both questions, and each share below is a share of one stress band within that group.

41.0% of the 2,967 responses with a stress rating of 7 or above gave one of the two shortest sleep answers, against 17.6% of the 2,386 with a rating of 4 or below: 2.33 times the rate. The 1,990 responses in the middle, at 5 or 6, sit at 30.9%, so the share climbs across all three bands rather than stepping once at the top.

Both answers come from one questionnaire, filled in on one day. Nobody here was followed from a stress rating to a sleep answer or back, so this does not show that stress shortens anyone's night, or that a short night raises anyone's stress rating. What it shows is that the two answers arrive together.

The Stress page reads the same pair from its own end, against the rest of what a high stress rating travels with.

Short sleep differs by four points across office, hybrid and home#

Four points separate the highest from the lowest. One point separates office from home. Hybrid is lowest in this edition.

This small spread does not establish that changing work arrangements would improve sleep. WHI 2025's work-arrangement table, on 602 home, 714 hybrid and 1,948 office responses, put the three modes in a different order, with hybrid highest and home lowest. No arrangement ranked lowest in both tables.

Nor can this survey say why any of the three sits where it does: it records the arrangement and the sleep band, not when people stopped working or how far they travelled. The work model page reads the same three cells alongside exercise and stress.

Natural short sleep exists, and it is rare#

"I'm a six-hours person" is offered as a fact about personality. It is a claim about physiology.

Genuine natural short sleep does exist, and it has been traced to specific variants. It is also rare. The first was found in a gene called DEC2, in a family with two affected carriers who slept about 6.25 hours a night against 8.06 for non-carriers (He et al., Science, 2009); that paper publishes no population prevalence at all. A second, in ADRB1, was identified in another short-sleeping family and is recorded in the Exome Aggregation Consortium database at about four per 100,000 (Shi et al., Neuron, 2019).

About four per 100,000 in one database is the recorded frequency of one variant, not a measure of how common natural short sleep is. What these family studies do establish is that a genuinely low sleep need has been traced to rare genetic variation — and nobody in this survey was genotyped, or asked to demonstrate one. A six-hour answer is not evidence of a six-hour requirement.