Last time I briefly mentioned administrative and routine data, this time we are going to dig a bit deeper on what that means.

Administrative (sometimes called routine) data refers to data collected by organisations as part of their day to day work. The nature of this data will vary greatly depending on the organisation collecting it. Let’s look at some example administrative data sets which are relevant to the kind of work we do in CHI.

GP data – Essentially everything your GP has ever written down (or more likely filled in on screen) exists in your GP record. This may also include information from a hospital specialist who has written to your GP to keep them in the loop about any ongoing treatment you are having. This is all essential information for the GP to have access to so when you visit them they can look back at your history to understand your current health needs.

Hospital data – When you visit a hospital, they will keep a record of why you attended and any scans or treatment they give you. This should be readily available if you visit the same hospital in the future. Generally speaking, hospitals cannot share data with other hospitals, this is why you may get asked the same questions about your medical background every time you go to a new hospital.

The Office of National Statistics (ONS) – The ONS collate and publish lots of data related to areas such as employment, births, deaths and the census.

Education data – The Department for Education collect a variety of data to e.g. monitor the performance of schools and sectors where pupils end up working.

Care home data – The Care Quality Commission collects data about care homes (and other organisations) to ensure that standards are maintained for their residents.

Crime data – The police collect information about crimes so they can better understand if there are people or places where more crime is committed.

There are plenty more of these data sets being generated as part of organisations’ routine work. In each case above you hopefully can see that it is necessary and reasonable for data to be collected and used as part of their daily activities.

Now just because these data sets exist doesn’t mean they are readily available to do analysis with, or even available at all. In each case the data controller (the organisation legally responsible for the data) has to decide if the data should be made available and under what terms. This could mean that we (as researchers) are only allowed to see e.g. the total count of people with asthma in each area. This might be all the data we need if we are just looking at broad trends in the region. Where more detailed data is needed it may be made available to a specific group of individuals who have done significant training to keep the data safe, it would be kept in a secure environment, and it would have had all the identifying information removed first. These steps (and more) are an integral part of how we keep data safe in modern data usage and we will cover these in more detail in a later blog post.

These administrative data sets may not be able to answer all of our questions, remember that the data is collected so an organisation can carry out its main purpose, so it is possible that relevant fields that we require to do our analysis were simply not collected, as they were not relevant to the original organisation. It is also possible that these data sets are missing people from them, so we have to be careful how we use them e.g. not everyone is registered at a GP practice, so we cannot assume that the population of an area is the same as the number of GP practice patients. Developing an understanding of these data sets and assessing if they can be used without introducing biases is what we spend a significant amount (perhaps a majority) of our time on when we work with them.

While these routine data sets in isolation are incredibly useful, sometimes we need to join them together to address difficult questions like does having a chronic illness as a child impact how well you do at school? Or does the level of employment in an area affect the standards in care homes? This is where the concept of data linkage comes in, which we will talk about next time.

– Dr Olly Butters, Care and Health Informatics theme.