Executive Summary

WHI 2025 read urban working India mainly through blood, alongside a 3,437-response behaviour survey. WHI 2026 keeps both and adds a third file: 74,339 hospital claim records. The survey is now 9,097 responses and the biomarker samples are roughly four times last year's. Glucose and low B12 shares remain close to their 2025 levels, reported stress is higher, and the claims file shows that covered family members account for most recorded incurred cost.

Biomarkers count reports, surveys count responses, and claims count records—not unique people. The datasets are not linked person by person. Laboratory results include insured employees and covered family, not a representative national workforce sample.

Elevated blood sugar is four times as common in reports from the 50s as in the 20s#

HbA1c is a single blood test that reflects roughly the preceding three months of blood glucose. This report counts a result at or above 5.7% as elevated and at or above 6.5% as being in the diabetic range, the same two cut points the US National Institute of Diabetes and Digestive and Kidney Diseases uses for prediabetes and diabetes. On that measure 38.7% of reports were elevated: 25.4% in the 5.7 to 6.4% band and 13.3% at 6.5% or above.

Four markers use the same cut points in both editions. Female anemia is included as each edition’s reported rate.

Four of the five moved by less than three points. Vitamin D deficiency is the exception and fell 5.8 points. Liver enzyme elevation rose the most of the four small movers, and low B12, at close to seven in ten reports, remains the most common finding in the table. The 2026 anemia rate counts female reports below 12 g/dL. WHI 2025 published its female rate with counts but no numeric haemoglobin cutoff, so the two rates are compared as reported rather than as a verified matched threshold. The detail is on the Blood Sugar, Vitamin B12, Vitamin D, Liver Health and Anemia pages, and in What Changed Since 2025.

Age separates the glucose result far more than the year does. Elevated HbA1c runs at 16.7% of the 2,050 reports from people in their 20s, 33.0% of 2,627 in the 30s, 52.9% of 1,001 in the 40s and 69.1% of 751 in the 50s. These are four age groups measured at one time, not one group followed through a career, and the spread is a reason to treat metabolic screening as a working-age question rather than an over-50s one. It is also specific to glucose: Vitamin D deficiency, for one, is worse in younger reports than in older ones.

Covered family account for a third of haemoglobin reports and two-thirds of claims#

Employees account for most haemoglobin reports, while covered family members account for most claim records.

In the biomarker file, 3,319 of the 9,585 haemoglobin reports belong to covered family members with no recorded employment. Four markers are cut by that split.

Three of the four run one way and B12 runs the other. The covered-family group mixes spouses, children and parents in one source category, so it is differently composed from the employee group rather than matched to it, and both ESR rates apply sex-specific thresholds within their respective groups. Self vs Dependent takes the split apart marker by marker.

The claims book is a separate file, with its own twelve months, and it runs the other way.

Non-employee relationships account for 66.5% of claims in total, a figure that includes two small residual groups outside the four rows above. The parent layer files fewer claims than the employee does but carries the largest share of the money, because its claims cost more on average than any of the other three displayed layers. Inside Loop's own commercial account tiers the parent layer's weight varies widely: it accounts for 42.7% of incurred amount in Enterprise-tier accounts and 18.8% in SME-tier accounts. Those tier names are account categories and carry no measured meaning about employer headcount or cover. See Employee vs Family Claims and Parent Claims, which sets out the tier cut in full.

The contrast concerns the mix of reports and claims in separate files. It does not measure who was offered screening or whether testing changed a later claim.

What fills the claim queue is not what carries the bill#

Loop's broker book recorded 74,339 claims over the twelve months to 26 December 2025. The report describes cost using shares and average amounts per claim, not aggregate rupee totals. Two facts govern how those measures should be read.

First, incurred is a recorded amount, not a payment. 12.8% of recorded incurred cost sits on claims that were still open at the snapshot, and the recorded deduction field is only a coded field. Nothing in the data measures what a household actually paid.

Second, the frequency ranking and the incurred-cost-share ranking are different lists. General health services is the most-filed category at 15.7% of claims, averaging ₹2,910 each. Musculoskeletal, injury and fractures has the largest incurred-cost share, 10.7%, on 5.2% of claims and an average of ₹89,455. Ranking every claim in the book, including those with no incurred amount recorded, the costliest tenth accounts for 47.6% of incurred cost. Top Claim Categories by Frequency and by Incurred Cost Share rank the two lists; Methodology explains the money fields.

Stress and short sleep rose; the regular-exercise share barely moved#

The stress comparison uses each edition’s published headline figures. The stress recall period and sleep answer bands need the qualifications below.

Stress is the largest movement in the table above: 33.9% to 40.4%, a 6.5-point difference between the published headlines. Other 2025 breakdowns imply different starting points; they also indicate higher stress in 2026 but change the size of the difference. What Changed Since 2025 gives those comparisons. The 2026 question asks about the last week. The 2025 page describes the item as daily pressure and does not state a recall period, so both figures use the same scale and cutoff, while the 2025 recall window remains unspecified.

The share reporting seven or more hours of sleep fell from 37.0% in 2025 to 30.7% in 2026.

The exercise result needs all three of its bands to make sense. Reporting no qualifying day rose 2.9 points, the one-to-two-day band fell 2.3 points, and the share reporting three or more days moved from 37.9% to 37.2%. The regular-exercise share is close to where it was, and the occasional band is the one that shrank. The three bands are shares of each year's own respondents, not a count of the same people moving between them.

The survey also shows differences by caffeine intake, work setting and industry. Short sleep rises at every step of workday caffeine, from 25.1% of the 1,684 responses reporting none to 41.9% of the 327 reporting four or more drinks. High stress is reported most often in the home-based group: 42.4% of 1,303 home-based responses, against 40.5% of 4,477 office-based and 38.5% of 1,565 hybrid. IT and software has the lowest short-sleep share among the six largest named industry groups, at 27.6% of 2,807 responses, against 38.8% of 482 in education technology and 33.0% of 630 in healthcare. Sleep, Stress, Movement & Exercise, Caffeine and Industries carry the rest.

Reported preventive testing rises with income, and it rises inside one job function too#

7,048 responses answered the preventive-care question, which offers five actions and a "None of these" box. 24.3% selected a full-body diagnostic test in the last year, and 42.3% selected "None of these". The five actions were a routine or preventive doctor consult, a full-body diagnostic test, a doctor consult for a symptom or illness, a company-organised health camp or screening, or a fitness assessment. Some responses selected both the None box and an action, so 42.3% is not a confirmed no-care rate. Split by income, the diagnostic-test answer moves a long way.

The share rises at every step of the ladder, though the first step is narrow, with the bottom two bands under a fifth of a point apart. The obvious objection is that income is standing in for occupation, that the bottom of the ladder is a different kind of job from the top. That objection does not account for all of it, because the same ladder appears inside one job function: within engineering and software development the share runs from 14.3% of 119 responses to 61.5% of 52, more than four times as high at the top as at the bottom. Those two end bands are small, and holding the job label constant does not hold age, employer or plan constant. Income & Job Role has the full cut.

Testing is self-reported. The income pattern also appeared in 2025, but its question and base differ, so the two editions' rates are not directly comparable.

Doctors lead as health-information sources; workplace programmes are near the bottom#

Asked where they get health information, 54.0% of 6,996 responses named a doctor or healthcare professional, a clear lead over everything else. The second most-selected answer is not a source at all: 24.8% selected "I don't actively seek health information", which is one option on a multi-select list rather than a verified absence of every other source. Then AI tools at 19.8% and Google search at 19.7%, a gap of two responses that should be read as level rather than as one overtaking the other, followed by family and friends at 19.2% and social-media influencers at 13.8%.

A workplace wellness programme is named by 8.0%, ahead of news and media at 7.4%. Health Information Sources breaks this down by age and income.

How to read the rest of this report#

The report’s 41 pages each state their main findings and scope. Risk Signals reads eight biomarker systems: blood sugar, lipids, anemia, vitamin D, vitamin B12, liver, inflammation and kidney. Lifestyle Drivers reads eleven behaviour topics, from sleep and stress to caffeine, alcohol, nutrition and GLP-1 awareness. Workforce Segments cuts the data eight ways: gender, age, work mode, income and job role, sedentary versus active work, employee versus dependent, parent claims, and Industries. Cost Signals, the major new section for 2026, is five pages on the claims book: who claims and what costs. Cities covers five metros, and the Employer Action Playbook turns the findings into plan-design moves. Methodology sets out how each dataset was built and what it can and cannot support.