Significance and impact of data quality on everyday hospital life, as well as control and decision-making processes

[1] Figure Information Quality Model DQIQ 2007   (in German)

What measures can be taken to achieve a high level of quality in the long term?

In order to permanently achieve a high level of quality, both initial and continuous measures and procedures should be taken. It is advisable to set the regularity to 24 hours and to automate it as much as possible, so that the risk of lack of continuity in case of staff changes and holiday replacements is as low as possible.

The procedure can be divided into the following areas:

• Data Profiling = Data Analysis
• Data cleaning = data cleansing
• Data monitoring = monitoring the data
• Process and procedure optimisation
• Use of IoT and suitable infrastructure (optional)

When the data in the database is incorrect

Often, hospitals have hand-crafted SQL queries on various topics such as the postcode and town, missing surnames, etc. These, however, only cover exactly one database.

However, these only cover exactly one database.

The approach per se is not bad, but there is no overview of what the data looks like in other systems. For example, has someone stored this data in the HIS (hospital information system) and is it only missing in the billing system?

What is needed is a system that includes the first three steps and offers interfaces for all systems. Ideally with a simple integrator tool. A system that can also be easily operated by trained staff and does not contain any complex programming of rules. Such systems already exist with and without IA (artificial intelligence) and are mainly used in the purchasing area of insurance companies, online trading and for checking the invoices of health insurance companies. In other regions such as the USA, Canada, Australia and Asia, such systems are already being used in full or in part in hospitals.

What are the effects of a poor quality level?

A direct or indirect reduction in revenue due to poor process flows, information flows and communication possibilities.

At the beginning there is the question of accessibility and workability.

First and foremost, the process flow is affected because, for example, corrections or alternative information procurement (telephone call instead of direct query in the system) cause a slowdown here. Weeks, months or even years pass before these are changed, because the effects are not directly in management, but at the lowest level of information flow aggregation, the „raw data“. This slowdown leads to indirect costs due to wasted time resources. Five minutes may not sound like much at first, but when you add up the process chain, you realise that inevitably others are affected as well,

The next point is freedom from errors.

If you always send the wrong data, this leads to a lack of credibility and decisions are not made on the basis of the data. Consequently, the process flow is also slowed down at this point as alternative information gathering is needed or worse, decisions are made on gut instinct.

This is supported by the question of use.

Can the data be used? Are they complete. Ask your controller this question sometime. The answer will be: if the data is in the system, yes.

After all, how can a controller know whether all the data is in the system and can be evaluated at the same time if he has no access to all the databases and process flows? A controller is only as good as its main database. It may find errors from other systems and have them cleaned up, but it has no control over whether important and relevant metadata is missing.

An example of this is the sending of laboratory services as a separate service, although these were provided on an outpatient basis and by an external laboratory.

In the hospital in question, I was able to have 4 of 12 million per year corrected in the long term through my S.ucin.o method. Retrospectively for 4 years and by means of a regular audit avoid future errors. This is because outpatient, external services are paid for in Switzerland by the party that provided them. In other words, the external laboratory. The controller had no chance here, because for him the data was correct. How is he supposed to recognise whether the service was really provided by his own or by another laboratory?

The last thing missing in this chain is the question of the addressee.

Are my reports, my views in the software structured in a comprehensible way?

Does the recipient know which data view is available to him? The structure of cockpits and reporting, as well as the visualisation of the process design, is a science in itself. A simple example is the date question. Are the services evaluated according to the date of discharge of all patients, according to the date of service or, for example, according to the period of time. All three options are correct, the question is, which one promotes the right decision?

Impact of unused data potential

A simple and practical example of wasted data potential and in this case wasted revenue is energy.

A 50-bed hospital in Switzerland (old people’s home or clinic) has costs of about 300 to 400,000 CHF per year here.

Very few hospitals have measuring stations at large facilities like the MRI or control their open-plan offices or operating theatres by means of heat sensors. Depending on the manufacturer, an MRI can save 8 – 10 single-family homes per hour. Incidentally, switching off for one hour has already been implemented in practice. What is important for medicine is that the MRI needs a maximum of 10 minutes to start up.

It is often too cold in open-plan offices because out of 10 possible seats, only 5 are occupied. If you regulate the temperature based on the maximum number of people and do not let the room temperature run at maximum cooling, you save staff resources indirectly through reduced sick leave and directly through reduced energy and heating costs for air conditioning.

Literature

[1] https://www.itwissen.info/Datenqualitaet-data-quality.html

Publication:

Tagungsband Hamek 2018, Herausgeber Jürgen Nippa

This article appeared as part of a keynote speech by our CEO Hannah Bock-Koltschin at Hamek 2018.