The health watchdog said on Tuesday it had launched an investigation into death rates at an NHS trust.

The Healthcare Commission said it was investigating Mid Staffordshire NHS Foundation Trust after data suggested its death rates "may be high" compared with other trusts.

The probe is focusing on patients admitted to the trust as emergencies. It will also investigate the quality of care provided across the trust, in particular of older people, after patients raised concerns.

"An apparently high rate of mortality does not necessarily mean there are problems with safety," said Nigel Ellis, the commission's head of investigations.

"It may be there are other factors here such as the way that information about patients is recorded by the trust."

The trust said it believed its rates were normal for a trust of its size and the population it serves.

It released figures showing its standardised mortality rate (SMR) was 127 in 2005/06. Nationally, the SMR is set at 100 so any figure above this reflects a higher than expected rate.

The trust said its SMR was down to 101 between May and October 2007, with an SMR for emergency admissions over that time period of 100.4.

The trust's chief executive, Martin Yeates, said it had worked with the Strategic Health Authority and concluded that the high rate was due to problems in the way patient information had been coded and recorded.

"We have, over the last year, employed more clinical coding experts to work within the various specialty departments to help staff to record information and improve the quality of data."

As a result, there had been a significant improvement in the trust's SMR.

The inquiry will also look at governance protocols for protecting the safety of patients at the two hospitals managed by the trust - Cannock Chase and Stafford.

The trust serves around 300,000 in people in Stafford, Cannock, Rugeley and surrounding areas of South Staffordshire.

The trust has set up a helpline for anyone with concerns - 08000 407060 - between 8am and 8pm Monday to Friday.