The numbers kept shifting. Which version reflected what was actually happening?
In Bangladesh, the agency entrusted with tracking a measles outbreak has revised its own figures three times in as many weeks — most recently cutting a single day's reported cases nearly in half within hours of publication. Such repeated corrections, whether born of fragmented systems or human error, do more than adjust a spreadsheet; they quietly erode the social contract between a public health institution and the people it is meant to protect. When the numbers that are supposed to guide a nation's response to disease cannot hold their shape overnight, the deeper question becomes not what the true toll is, but whether the machinery for knowing it still functions.
- Bangladesh's DGHS posted measles figures showing nearly 2,000 suspected cases in a single day — then replaced them with roughly half that number just two hours later.
- The revision was not a minor tweak: entire divisions appeared to have swapped data, with Mymensingh's count collapsing from over 1,000 cases to just 44, while Sylhet gained the exact figures Barishal had lost.
- This is the third major data upheaval in two weeks — following the late addition of 46 deaths in May and the removal of over 4,000 duplicate cases from Rajshahi shortly after.
- Public health experts warn the pattern is actively corroding public trust, leaving citizens unable to gauge the true scale of an outbreak that has already claimed lives.
- Authorities have issued show-cause notices to hospital and divisional officials and promised a fresh investigation, but each new pledge of accountability arrives in the shadow of the last unresolved one.
On the afternoon of May 22, Bangladesh's Directorate General of Health Services published measles figures showing 1,967 suspected cases and 1,878 hospital admissions in a single day — the highest counts since mid-March. Two hours later, both numbers had been cut roughly in half: 972 cases and 883 admissions.
The revision was not a rounding correction. Examined side by side, the two releases told entirely different stories about entire divisions. Mymensingh, credited with over 1,000 suspected cases in the first version, appeared in the second with just 44. Barishal lost 101 cases — and Sylhet gained almost exactly that many, inheriting the precise figures Barishal had shed. The DGHS attributed the change to updated information from the Mymensingh Civil Surgeon Office, though the symmetry of the swapped numbers pointed more toward a data entry error than a genuine epidemiological development.
It was the third major revision in as many weeks. On May 10, the directorate had quietly added 46 suspected deaths absent from earlier counts. On May 18, it removed 4,340 cases, blaming duplicate submissions from Rajshahi Medical College Hospital. Now a third upheaval had arrived before the dust from the second had settled.
Public health expert Mushtuq Husain said the accumulating corrections were undermining confidence in the official record. The question he and others were raising was not merely technical — it was whether any version of the data reflected what was actually happening across the country.
Prof Zahid Raihan, the DGHS additional director general for administration, acknowledged the latest mismatch and confirmed that show-cause notices had been issued to officials in Rangpur and Rajshahi for submitting incorrect information. He promised an investigation into the May 22 revision as well. The death figures, at least, remained unchanged across both versions of that day's release.
But for a public trying to understand the scale of a live outbreak, the fog of repeated corrections carried its own cost. Each revision implied either a system too fragmented to produce reliable numbers or an institution too careless to catch errors before publication. Whether the promised investigation would restore credibility — or whether another revision would arrive first — remained an open question.
The Directorate General of Health Services posted its measles figures on the afternoon of May 22, then took them down and replaced them with different numbers two hours later. The first set showed 1,967 suspected cases in the previous 24 hours and 1,878 hospital admissions—the highest figures since mid-March. The revised version, released shortly after, cut both numbers roughly in half: 972 suspected cases and 883 admissions.
The discrepancy was not a rounding error or a minor adjustment. When the numbers were examined side by side, entire divisions had been reshuffled. Mymensingh, which the first release credited with 1,039 suspected cases and 1,024 admissions, appeared in the revision with only 44 cases and 29 admissions. Barishal dropped from 144 cases and 150 admissions to 43 and 43. Sylhet, meanwhile, climbed from 43 cases and 43 admissions to 144 and 150—the exact figures Barishal had lost. The DGHS attributed the revision to updated information from the Mymensingh Civil Surgeon Office, but the pattern suggested something more like a data entry correction than a genuine epidemiological update.
This was the third major revision in as many weeks. On May 10, the directorate had added 46 suspected deaths that had not appeared in earlier counts, citing a review of records from three medical college hospitals and regional civil surgeon offices. A week later, on May 18, it removed 4,340 suspected cases from the database, explaining that Rajshahi Medical College Hospital had submitted duplicate information. Now, on May 23, came the third upheaval.
Mushtuq Husain, a public health expert, said the pattern was corroding confidence in the official account. "The authorities should place greater emphasis on the matter to avoid such revisions," he told a correspondent. The repeated changes raised a fundamental question: if the numbers kept shifting, which version—if any—reflected what was actually happening on the ground?
When contacted, Prof Zahid Raihan, the additional director general for administration at DGHS, acknowledged that yesterday's data had been revised because of overlapping information from Mymensingh Medical College Hospital. He added that the directorate had issued show-cause notices to the director of Rangpur Medical College Hospital and the Rajshahi divisional health director for submitting incorrect information. He also said an investigation would be launched into the latest mismatch, with action taken against anyone found responsible.
The death figures, at least, remained consistent across both versions of May 23's data. But for a public trying to understand the scale of a measles outbreak affecting thousands of people, the repeated revisions created a fog. Each correction suggested either incompetence in data collection or a system too fragmented to produce reliable numbers. Either way, the credibility of the institution tasked with tracking the outbreak had been damaged. The question now was whether the investigation Raihan promised would restore it, or whether the next revision would arrive before the public had time to trust the current one.
Notable Quotes
The frequent changes in the data cast doubt in people's minds about the authenticity of the information. The authorities should place greater emphasis on the matter to avoid such revisions.— Mushtuq Husain, public health expert
Today's data was revised due to overlapping information from Mymensingh Medical College Hospital.— Prof Zahid Raihan, additional director general (administration) of DGHS
The Hearth Conversation Another angle on the story
Why does it matter that the numbers changed? Isn't some revision normal when you're tracking an active outbreak?
Revision is normal. But this wasn't a small correction—they cut the case count nearly in half in two hours. And the pattern is strange: cases disappeared from one division and appeared in another, like someone moved numbers around rather than recounted them.
Could it have been an honest mistake?
Possibly. But this is the third major revision in two weeks. May 10, May 18, May 23. At some point, the pattern itself becomes the story. People start asking whether anyone knows what's actually happening.
What did the health directorate say when asked about it?
They blamed overlapping data from Mymensingh Medical College Hospital. They said they'd issued notices to other hospitals for sending wrong information. They promised an investigation. But they didn't explain why the numbers moved between divisions the way they did.
Does that explanation make sense to you?
Not entirely. Overlapping data would mean you have duplicates to remove. But these numbers didn't just disappear—they reappeared somewhere else. That suggests either a data entry error or something more systematic.
What's the real cost here?
Trust. If you're a doctor trying to understand how severe this outbreak is, or a parent trying to decide whether to vaccinate, you're looking at numbers that keep changing. You don't know which version to believe. That uncertainty is dangerous in a public health crisis.