In this part 1 of a 2-part blog – I am going to look at some of the key considerations in how providers can look to get more value from their data – both for their patients, and for patients further afield.

Taking the Hippocratic Oath is a right of passage for any clinician as they move from training into clinical practice. The ethos and mandate to do no harm, to respect the patient and their confidentiality, and to use the knowledge gained and passed on to doctors to treat the sick, is still paramount to the vocation of clinical medicine. The world we live in now is obviously very different to that of the original oath, but those founding principals have not waivered.

One of the main differences in today’s healthcare world is just how connected it is - with disease discovered and treated in one part of the world shared and impacting the health of others thousands of miles away. Now coined Real World Data – the information on how patients present with disease, their social circumstances, how they are treated and how they recover or progress (or anywhere in between) is hugely important in informing how we treat other patients with similar illnesses. The rapid expansion of healthcare data in electronic format, due in large part to the adoption of Electronic Health Records (EHRs), means that in the healthcare ecosystem, there is potentially a goldmine of data waiting to be used alongside the stethoscope as a key part of the clinician’s tool kit.

To ensure the data is prepped and ready to use appropriately, there are some key considerations that need to be taken into account. In this blog – I will outline what I see as the most important.

Keep Patient Information Confidential

One of the biggest and most legitimate concerns is upholding the privacy of patient information. There is an enormous amount of harm that could occur with these semi-regular data breaches. In 2015, one institution had an IT incident that breached the data of close to 79 million. It should be obvious that when moving that data out of the EMR it is to everyone’s advantage to remove identifiers but preserve integrity of the data if at all possible. Statistical anonymization to GDPR and HIPAA standards is a key step in the process of data transformation.

The original Hippocratic oath stated bluntly the importance of patient confidentiality: “And whatsoever I shall see or hear in the course of my profession… I will never divulge, holding such things to be holy secrets.” The sanctity of confidentiality is one of the most, if not the most, important pillars of medicine. Only by knowing that the information shared between patient and doctor is treated with confidentiality, will patients tell the full story of their health and illness. But does this mean that the insights gleaned from one patient should not be used to the benefit of others? In my opinion – certainly (and practicably) not. In fact, it is from our experience as clinicians that we learn and improve how we help others. The signs we missed are those we will never miss again. Of course – when taking learnings from one patient to the next – it is paramount to preserve the confidentiality of the first patient. To tell a patient that her symptoms were the same as her neighbor’s would be breaking the oath. Nonetheless, good clinical practice is underpinned by continuing professional education – no better realized than in the field. Therefore – where insights or data gleaned in a healthcare organization are going to be used outside that setting – it is essential that they are done so within established privacy regulations.

Don’t ignore the unstructured data!

While structured data in healthcare is informative and useful for certain decisions – there is a huge amount of information about patients that remains unaccounted for if we only look at billing and administrative codes. Signs, symptoms, and indications of disease severity are largely unaccounted for in traditional structured data – yet are vitally important in clinical decision making. Take oncology for instance – structured data will tell us that a patient has breast cancer – but only the unstructured data will reveal the histology, tumor size and staging. Making the right treatment decision is impossible without these additional insights. Technologies such as Natural Language Processing that can surface this information – and more importantly – normalize it to preferred terms (e.g. Breast Cancer = Breast Ca = Carcinoma of breast) are therefore key in health data transformation.

Aim for industry standard data models for better sharing of knowledge

Knowledge sharing is by its definition collaborative – and therefore organizations looking to make their data more clinically valuable should do so in a way that enables other organizations to benefit from shared insights. Therefore – representing your clinical data in a standard way – such as a common data model (CDM)– is important. One such example is the Observational Medical Outcomes Partnership (OMOP) CDM, a partnership based on sharing data sources from individual healthcare institutions and is the effort of the opensource group Observational Health Data Sciences and Informatics (OHDSI). This CDM seeks to harmonize disparate coding systems and terminologies to enable search across data ecosystems to advance clinical research and quality of care. While conversion of data to such models can be done manually – this is another area where technology can be used to drive the transformation of healthcare data.

We cannot and should not go back and withhold ourselves to the original Hippocratic Oath itself or even those antiquated paper medical charts. After all, a lot has changed since 400 BC Greece and some of the advice in this antiquated oath is highly controversial, but the original purpose of it should remain today. Withhold yourself to the highest moral conduct, share the knowledge and protect the patients and how do we do that today in this modern world? Data transformation.