But the bureau urges caution, saying numbers are small and it will be watching closely to see whether the increased deaths continued into April.

And Professor of infectious diseases at the Australian National University Peter Collignon said it was more likely that people with pneumonia and diabetes didn't get medical help in time because of the lockdown.

The data shows a jump in deaths attributed to pneumonia and diabetes, in particular, at the end of March - both strongly associated COVID-19.

Ninety-nine deaths were attributed to diabetes in the last week of March - 30 per cent higher than the average for that week over the past five years, of 76 deaths.

A big jump was also seen in deaths attributed to influenza and pneumonia, with 76 deaths attributed to those diseases in the final week of March this year, compared with a five-year average of 42.

Bureau director of health and vital statistics James Eynstone-Hinkins said the increase was all in pneumonia deaths, with influenza deaths actually lower all year.

"At the end of March we can see there are higher numbers, especially for pneumonia and diabetes, so obviously that is of interest," Mr Eynstone-Hinkins said.

"We can see an increase. We're paying close attention."

But he urged caution, saying numbers varied month to month and there were "lots of things that happened in the last week of March that could have impacted on counts of death", including the lockdown itself.

"We really do need to think about all these different scenarios before we draw conclusions," he said.

The bureau analysed deaths certified by doctors - 85 to 90 per cent of all deaths - rather than those which go to a coroner for investigation, such as accidents.

The cause of death is the cause cited by the doctor.

Overseas, large numbers of "excess" deaths in a number of countries overseas have been seen as indicating an underlying level of undetected coronavirus in the community.

Mr Eynstone-Hinkins said in Australia, the number of coronavirus deaths was small, as was the number of deaths from other individual conditions, varying week to week.

"We're right down in the detail looking at very small numbers by different diseases and conditions. We need to be careful over-interpreting these small numbers," he said.

Professor Collignon said if coronavirus cases had been missed, the excess deaths should be much higher among the elderly, since coronavirus kills 80 year-olds at 100 times the rate of 30-year-olds. The extra deaths occurred mainly among 65 to 84-year-olds, and to a lesser extent in 45 to 64-year-olds, but not in over 85-year-olds.

It was more likely the excess deaths were a result of people not seeking treatment for other conditions at the height of the lockdown, especially as it was difficult to get face to face medical appointments, he said.

Diabetes and pneumonia were both chronic conditions vulnerable to acute deterioration if not treated - diabetes if sugar levels get out of control and pneumonia without antibiotics.

"If they're not seen promptly, that's where I expect you might see more deaths due to the delay in getting testament," he said.

The question could be settled by testing any existing blood samples of people who had died, he said.

The Australian Medical Association said it was too early to jump to conclusions but spikes in deaths were expected.

"We know based on past pandemics that people avoid seeking care for acute and chronic conditions during a pandemic. We also know from various sources that many people did defer medical care during the peak COVID-19 period," the association said, urging people to keep seeing their GP and going to hospital.

Overall, 2649 deaths were recorded in the last week of March, compared with the average of the previous five years of 2526 deaths in that week. That is an increase of 5 per cent, an increase not unexpected in light of population increase.

By state, NSW recorded no increase is seen in deaths, with increases in Victoria, Queensland and Western Australia. ACT and Northern Territory deaths are counted together and show no increase, and in fact a fall in the last week of March.

Nationally, deaths attributed to dementia have tracked higher all year, well above the average every week, and peaking in the final week of March at 300 deaths that week, 24 per cent higher than the five-year average. But Mr Eynstone-Hinkins said dementia deaths had been steadily increasing in recent years.

COVID-19 is listed as the cause of death in 89 registered deaths by the end of May. Most had acute respiratory symptoms, including pneumonia, the bureau said.

Of deaths attributed to COVID-19, the data confirms that deaths largely occur among people with underlying health conditions.

More than two-thirds (69 per cent) had pre-existing conditions. The most common was high-blood pressure, followed by dementia, diabetes, heart disease and cancer. The heavy impact of pre-existing conditions has already been reported in health department data.