Three provincial North Island hospitals have unexpectedly high death rates, according to data released to the Herald after a two-year wait.

Whangarei, Whakatane and Gisborne Hospitals show up as "outliers" needing special attention in the death data compiled by the Ministry of Health and the Health Quality and Safety Commission.

The data is being published on the commission's website today. The two agencies gave it to the Herald before its public release after the paper's complaint to the Ombudsman under the Official Information Act.

The data compares the actual number of deaths with the number "predicted" based on the risk profile of a hospital's patients, including age, diagnoses, prior illnesses and socio-economic deprivation. The data covers inpatients who died within 30 days of admission, including those who died after they left hospital.

In 2013, the latest year for which the data is available, Whangarei Hospital had 439 deaths, 16 more than the predicted number of 423.

The other two outliers' mortality figures are: Whakatane, 173 deaths - 17 more than the predicted 156; and Gisborne, 176 - 35 more than the predicted 141.

Nationally, the ratio of actual to predicted deaths declined from 2007 to 2013.

There is international debate about the usefulness of this kind of data, called hospital standardised mortality ratios or HSMRs.

Auckland District Health Board, whose ratio is below the national average, considers the data to be a "key" measure of "quality and safety", according to its Health Needs Assessment report.

"For Auckland Hospital, [the ratio] has remained stable for the past four years at around the average level for comparable hospitals. For Starship hospital, the HSMR level is stable but sits 25 per cent higher than the average for other hospitals. This may be due to its relatively small size and the high complexity of the patients."

The ministry says the ratios should not be used to compare hospitals, to calculate "avoidable deaths" or to draw "simplistic conclusions about quality of care". But they can be a warning of potential problems and the need for investigation if the HSMR suddenly increases or is outside the statistical "control limits", such as for the three outliers.

The three hospitals say they provide safe or good-quality care and their results are higher in part because of the inclusion of palliative care in the data. Their centres have poorer access to hospice care - meaning higher proportions die in hospital or soon after - and terminally ill patients are not always recorded as such.

Gisborne and Whangarei say other factors which may not be accurately reflected in the data are their populations' high levels of deprivation and multiple illnesses.

Northland DHB chief medical officer Dr Mike Roberts said Whangarei's result was out of line with the low and stable findings under a previous assessment method and did not reflect the hospital's average or better performance on several direct measures of clinical quality.

"We have been working with the ministry to see if we can understand why this result is so different, and why it particularly disadvantages a few DHBs with similar demographic challenges and extremely high-need populations."