The government's flagship initiative on childhood obesity has run into "practical difficulties" and opposition from parents, ministers admitted today.

Schools in England have measured less than half the children they were supposed to. As a result, the National Childhood Obesity Database has failed to generate sufficiently accurate data to enable schools and health professionals to tackle the problem.

A report published by the Department of Health warns that the figures collected in the first year of the scheme "systematically underestimate" the prevalence of overweight and obese children and must be treated with "considerable caution".

If, as the department hopes, more accurate data is collected this year, it will look as if there has been a big jump in obesity among children.

The admission comes in the same week that the Department for Education and Skills was branded "unfit for purpose" in a report from the National Audit Office, which found that 12 out of its 14 key targets were likely to be unreliable.

The largest database of its kind in the world, the NCOD was set up by the government as a tool with enormous potential for tracking and analysing trends in childhood obesity, and tackling this major public health problem.

Parents were asked for their consent to having their children measured, and the report said: "There is anecdotal evidence of higher rates of opting out of the measurement process among heavier children, which is supported by the findings of this analysis. This means the figures obtained from the NCOD are likely systematically to underestimate the prevalence of overweight and obesity."

The report also said there was confusion over which primary care trusts were responsible for which schools.

Although 80% of primary care trusts returned some data on schools in their area, response rates varied widely across England, undermining the usefulness of the data.

Altogether, 538,400 children in reception year and Year 6 were measured - approximately 48% of those eligible.

The Department of Health said nationally, of those children measured, 12.3% of girls and 13.4% of boys in Year R were found to be overweight, and 9.2% of boys and 10.7% of girls in the same year group were found to be obese. In Year 6, 13.8% of boys and girls were overweight, and 15.4% of girls and 18.9% of boys were obese.

"Comparing the NCOD results for different areas shows that the higher the response rate, the higher the apparent prevalence of obesity. This suggests that as response rates increase the estimates from the data more closely approach the true prevalence for that area," said the report's authors from the South East Public Health Observatory.

They warned: "It is therefore likely that the more accurate data anticipated in 2006-07 will appear to show an increase in obesity prevalence."