Amitava Banerjee: Is conflict of interest a concern in healthcare IT?

In the UK, electronic health records (EHR) have been almost universal in general practice since the 1990s, and were deployed across hospitals in the early 2000s. The Professional Records Standards Body (PRSB) was set up in 2013 “to develop high quality, consistent care records and promote their use.” As I sat in the third annual general meeting of the PRSB recently, I wondered why standards had taken so long to become a priority. As a clinical academic in health informatics (HI), two issues struck me. Firstly, quality and standards in healthcare information technology (HIT), require knowledge and engagement on the part of patients. The PRSB is leading in this respect, but sadly this is not the norm in HI, whether research or clinical practice. Secondly, conflict of interest (COI) is barely mentioned in HIT, unlike other sectors such as pharma and medical devices, where it is acknowledged that bias in evidence impacts research, guidelines, medical education, and care. A plethora of apps, EHRs, and other technologies are sold direct-to-consumer, from patients and carers to the public and clinicians. Without unbiased evidence, these consumers cannot make informed decisions. Moreover, most apps and EHRs are not considered as medical devices by regulatory authorities such as NICE (National Institute for Health and Care Excellence) and MHRA (Medicines and Healthcare Products Regulatory Agency). Patient safety cannot play second fiddle to COI.

Implementing HIT on a large scale has been problematic, particularly in the UK. However, renewed enthusiasm has come from high-level recommendations and national initiatives, such as the Global Digital Exemplar Programme, NHS Digital Academy, and the Faculty of Clinical Informatics. Moreover, the Wannacry cyber-attack in May 2017 showed that robust HIT infrastructure is a pre-requisite of data security. Globally, $80 billion is spent every year on HIT, including EHRs, telehealth, electronic ordering systems and clinical decision support systems, with 40% predicted growth by 2019.

Given the scale of investment, the relative lack of a transparent COI policy in HIT seems against the tide, with far-reaching consequences in both clinical and academic spheres. There are implications for industry-funded education and training of health professionals, data scientists, managers, and researchers. Criteria for procurement and evaluation are neither agreed nor standardised. Ironically, the decision to use one HIT solution over another is usually not systematically based on evidence or data, and is open to COI at the level of hospital boards as well as individual clinicians. There is no cost-effectiveness methodology for HIT. Unlike other aspects of healthcare, there are no tariffs or consensus guidelines. Paradoxically, albeit inadvertently, variations in care may be encouraged rather than reduced without a coherent HIT strategy. At best, public money could be spent with greater accountability, but our ability to measure quality improvement is hampered by deficiencies in metrics, research, and policy. At worst, there may be patient harm if technology is used when a safer alternative exists, e.g. pharmacist-led HIT interventions reduce medication errors.

The prevailing view of HIT as infrastructural rather than integral to patient management is outdated. There are definitely components of HIT, such as e-prescribing, decision support tools, or telehealth, which are interventions. Trials of one EHR component or system versus another (e.g. cross-company) are rare and the same is true across HIT, from wearables to use of apps by clinicians and patients. If current methods of evaluation do not fit HIT, then new models can and must be developed. HIT is a growing and evolving discipline. Genomics, decision science and advanced data analytics, including machine learning, are examples of fields where there is increasing industry, academic and clinical interest. However, both research and patient care will suffer if COIs are not appropriately managed.

Three major challenges persist if COIs place commercial benefit above system solutions. Firstly, data linkage is hampered along the patient journey from primary to tertiary care and across health and social services because each step is delivered by a different company with a different interface. Secondly, providers are neither incentivised nor mandated to consider interoperability, despite existing clinical ontologies and coding systems, which makes procurement “all or nothing” (e.g. a whole hospital’s system must be changed). Thirdly, hospitals wanting to add functionality to their HIT are often penalised and must wait until the end of their contract. Large, multinational companies are not incentivised to produce locally tailored solutions and hospitals are dis-incentivised to develop their own solutions, as evidenced by the low number of Trusts with locally designed EHR in the UK.

COI is not the only problem with implementation and evaluation of HIT. However, to fulfil its promise of enhanced financial, clinical, and scientific efficiency, fundamental changes are necessary. There is a rich experience and literature documenting the association between COI and harms to patients, populations and research. Guidelines for COI relating to HIT can help industry and health systems to improve both science and care in a field with enormous potential.

Amitava Banerjee is senior clinical lecturer in Clinical Data Science and Honorary Consultant Cardiologist at University College London. He is particularly interested the application of health informatics to improving patient care and is active in research, teaching and clinical practice.  

Competing interests: None declared.