A NSW Bureau of Health Information report published on Wednesday compared mortality and readmission rates in 2012-2015 with 2009-2012.

It found improvements in mortality in all seven conditions analysed, including heart attack, stroke, pneumonia, hip fracture, heart failure and lung disease, while readmission rates reduced for four of eight conditions.

But more hospitals performed below expectations when it came to the number of people who died from lung disease and pneumonia since the last time the data was assessed.

NSW is the only state in Australia to publish mortality and readmission rates at individual hospitals, allowing trends in the system to be identified and hospitals to evaluate their own performance.

Professor Philip Clarke, from the University of Melbourne Centre for Health Policy, said the most recent mortality data he could find at a Victorian hospital dated back to 1859.

"People often have the view that somehow hospitals were better a decade ago," Professor Clarke said.

"But what this shows in terms of hard outcomes is actually things are improving over time.

"How does NSW hospitals compare with the rest of Australia? Unfortunately no other state regularly publishes comparable information to enable such comparisons, so it is impossible to say."

Compared to other countries NSW performed similarly for most conditions, but had a higher mortality rate for ischemic stroke and a relatively low mortality rate for heart attack.

Professor Brad Frankum, president of Australian Medical Association NSW, said mortality rates were falling due to better treatment and prevention measures, with cholesterol medication lowering the incidence of heart attack and stroke units in hospitals allowing doctors to respond faster.

It was not clear from the data why some hospitals performed better than others, but the identification of outliers would allow those hospitals to investigate their practices, he said.

Professor Frankum did not support the publication of the readmission and mortality rates for individual doctors because it would allow the public and media to compile league tables without taking into account the risk profile of the patients that those doctors treated.

"I do think as practitioners we should be doing it ourselves," Professor Frankum said.

"I would be much more interested in a system where part of our registration included auditing our outcomes and comparing that to other practitioners."

This would ensure that a profession renowned for evidence-based practice could apply the same scrutiny to its own performance.

"I do think the evidence base needs to increase," he said.

Readmissions rates improved most notably for heart attack and total hip replacement, but became more frequent in the case of ischemic stroke and pneumonia.

Prince of Wales and St Vincents had better than expected mortality rates for three conditions - heart attack, coronary heart failure and pneumonia in the case of Prince of Wales, and coronary heart failure, pneumonia and hip fracture for St Vincents.

Four hospitals - John Hunter, Port Macquarie, Manning and Tamworth - had lower than expected mortality rates for three conditions.