I’m a hospital-based medical specialist. One of the things that annoys patients (and healthcare workers, too) is the rigid siloing of medical problems – “I’m not here about your diabetes, I’m the heart specialist”.
The siloing of expertise isn’t just an issue for medical specialists, though. Continuing the health example is the seemingly endless debate about the “best” form for the electronic medical record. The fundamental issue is that the “essential requirements” of a national electronic health record are a spectrum, between extremes resembling “universally accessible (from any device by any healthcare worker), single-format cradle-to-grave health information” and “completely granular, user-controlled information, encrypted with an unbreakable algorithm, only accessible by registered professionals on an authorised device”.
The “correct” position on this spectrum is of course both ends, depending on whether you are a healthcare worker who goes to a training session in the library to learn how to use Microsoft Word (these people exist, and look after you while you are in hospital) or a tinfoil-hat wearing privacy advocate.
This analogy applies to most things a government needs to make decisions about (except, it seems, marriage equality, for which the current federal government seems pretty keen on handballing responsibility back to the people).
It is the role of government (with some frank and fearless advice) to set the position of this slider on the spectrum. This, by necessity, means that some people will be unhappy – we are a plural society, and not everyone has the same opinion. We have had some pretty solid disagreements in the Future Wise bunker, despite our common attitudes.
Marriage equality is a challenging example, because the slider is a switch – let gay folk get married or not. The other problem is that (even if there were a possible middle ground), advocates for each option are highly polarised and disinclined to cede ground to the other side.
The same sort of balkanisation occurs with other health / tech issues. I’ve previously done war with some geeks and journalists who were convinced that NBN-enabled telehealth would result in massive numbers of discharges from hospitals and stupendously large cost-savings (pro-tip: it wouldn’t). Most health workers don’t really understand the technological limitations (I have been in electronic-health meetings where doctors say things like “well, just double the bandwidth, then” to solve the problem of slow program load times) and the geek-set don’t really grok how hospitals work (see my blog post re: “patients can be monitored at home“).
Which brings us to the 2016 Census.
By way of background, the Australian Census is a 5-yearly statistical survey of the Australian population, carried out by the Australian Bureau of Statistics. It has always been done on paper, with small armies of census staff knocking on doors and handing out and collecting forms. The census includes lots of information on demographics, employment, income and various other statistical gems. ABS has always provided excellent service, and the dataset is both rich and useful.
In 2012, the ABS announced that it would be retaining the names and addresses of a small subset of census samples. There was also an option to opt-in to the “Census Time Capsule” so that all data would be retained, to be unlocked in 99 years as a historical exercise.
In December 2015, ABS announced that all names and addresses would be retained for four years, to allow for data linkage between various government datasets, to improve the statistical picture of Australians. It was done without visible public consultation (the submission window was three weeks in November, there were three public responses – all opposed), included a privacy impact assessment which considered the risk of a loss of public trust in the census and decreased participation as “very low”. And then went ahead anyway.
The Bureau’s comms on the issue was abysmal, and I’m sure will become a case-study in how to not do corporate disaster management from beginning to end (also not the point of this post).
I tweeted my thoughts about the risks of retention of potentially identifiable census data. By and large, they are the same as the risks we have previously discussed relating to the mandatory retention of telecommunications data – which, although it is directly identifiable, does not include the richness of details like income, religion and all the rest of it which the census does.
Unsurprisingly, there were two(ish) groups of responses – endorsement from the tech community, and quite a number of researchers and healthcare workers bollocking me for potentially threatening the rich research dataset that the census provides, which went along the lines of this letter from Professor Fiona Stanley to Crikey.
Who is right? Well, of course, they both are.
Data linkage, as Prof Stanley points out, has amazing potential for public health research. Whether the privacy-aware like it or not, we live in the time of Big Data, and there certainly have been major advances made on the back of this sort of data. However, medical research works on the principle of consent. Research subjects have the option of opting out of this sort of data linkage if they so choose, which they do not with the Census.
What has been (largely) missing from the epidemiology professional commentary is awareness of the technical nature of the risks. Even with the most secure data management procedures, there is always the risk of internal data breaches. This has been seen with police data again and again and again. As there is so much more highly personal information in the census data, the security bar should be set even higher than for general research data.
This nexus is where Future Wise sees its role – approaching complex issues not from any one particular point of view, but trying to integrate multiple competing priorities, and ensuring the end result as as good as possible for as many as possible. We’d love you to help us – join the discussion on our forum. I will be sending some questions out to various people in privacy and public health research for further comment, and will post their responses on the forum.
Thanks to Justin Warren for picking up a couple of minor boo-boos in the first version of this post.