Without Improved Data Quality Our Health Systems Will Fail

Without Improved Data Quality Our Health Systems Will Fail

The act of gathering large amounts of aggregated medical record data into one place creates a transparency that we haven’t had before.  It represents the next step in a maturity journey, allowing us to shine a light on some of the real root causes of systemic quality and safety issues in our health systems.  The fact that we are reaching this maturity stage should be welcomed


With the explosion in medical knowledge that I’ve talked about in previous posts, it is clear that we increasingly need to arm clinicians with better clinical decision support, ensuring that appropriate medical knowledge is available at the point of care.

However, the automated application of this knowledge (where appropriate) is reliant upon high quality, structured, atomic, coded data.  This requires data that isn’t just of high structural quality, but has high clinical quality.  You see, data can be structured, atomic and coded, and still be crap.  Ultimately, to be of high quality, clinical data also needs to be a true, accurate representation of reality.

During a recent consulting engagement, I had the opportunity to spend some time looking at and thinking about data quality within the Australian National eHealth Record, known as the My Health Record.  As part of this project, we looked at a number of de-identified clinically-authored CDA documents.

Without sharing more information than I should, it would be fair to say that the quality of clinical content within Shared Health Summaries (curated by General Practitioners) and Discharge Summaries (sent from hospitals) was often poor.  Examples of quality issues included:

  • Contra-indicated medications in current medications lists (i.e. medications that shouldn’t be taken together because of a significant risk of Adverse Drug Events occurring)
  • Medications being taken for conditions that were not in the problem list
  • A large proportion of patients on > 10 medications, where good practice suggests a strong clinical case for consolidating / reducing medications


The prevalence of these clinical data quality issues points to two possible causes:

  1. Systemic issues in the quality of medical record keeping
  2. Issues with the quality of medical care being delivered


Whilst both of these causes are undoubtedly issues, I’d like to focus on the first one.

In a previous post on patient-curated medication lists I explored one of the reasons why clinicians often don’t do a great job of data quality in medical record keeping:

“By contrast, clinicians are extrinsically motivated, i.e. they are paid to curate patient health records.  As I’ve discussed in previous blog posts, intrinsic motivation is generally a stronger lever for behavioural change than extrinsic motivation.  Consequently, it should come as no great surprise that busy clinicians, who are paid under a fee-for-service model for appointments, but not explicitly (in Australia at least) paid for data quality, often don’t do a great job of this.  Now of course there are exceptions to this.  But I don’t want that to get in the way of my key observation – patients have a stronger motivation to capture their data correctly than clinicians do.”

Consequently, if we are going to see an improvement in clinical data quality, we need to involve the patient.  We need not only to share clinical information with patients and their caregivers, but also to give them the opportunity to act as another line of defence against error, by flagging and correcting clinical data where it is wrong or misleading.  This is arguably the most powerful mechanism available to us in driving improvements in clinical data quality.

In addition to this, we need to find new, stronger motivations for clinicians to do a better job of medical record keeping.  This starts with stressing the importance of data quality during the process of medical education – something that I’m not convinced that we are doing well.  However, this goes deeper.  Given the pivotal role of data quality in our future health systems, we have to start seeing data quality as synonymous with healthcare quality.  You can’t have one without the other.  Without improved data quality our health systems will fail.


One thought on “Without Improved Data Quality Our Health Systems Will Fail

  1. We don’t do data quality for the sake of data quality, we do it to improve, support or enhance some aspect of our lives. If physicians and providers want to reduce the costs of liability, accidentally prescribing wrong or fatal mismatches…etc, they should consider using the Rx data described in this article. If they do the quality errors will force their attention toward quality and therefore the systems and processes that ensure quality. If a physician knows that lives are dependent on the coding, they are pretty good at ensuring high quality. Funding for data management is unavailable, until the data is used- and its improvement is required. My guess is that the Rx data discussed in the article isn’t used currently therefore has little need to be corrected.

    It’s worthless to collect patient feedback. No, it’s detrimental to your public image and maybe liability, if you ask them to correct the error and you system/process doesn’t do it. Especially if lives are at risk. So my recommendation is to ensure the feedback loop works first–regarding prescription data quality, I know it doesn’t at a few medical providers that I’ve been involved with. So I gave up trying to correct that Rx list.

    The point is, that data quality is just what professionals like us do, AFTER the business change has been implemented: be it dependence on the data, or innovation using the data…without the business driven spark there will be no data quality fire.

Leave a Reply

Your email address will not be published.