A recently published research piece published in the British Medical Journal has raised questions about an an approach that is widely lauded and promoted by its protagonists. A study by a Canadian group examined improvements achieved with clinical decision support systems, and the findings should be unsettling for funders committing huge sums of money to the broad implementation of computerised clinical decision support. The researchers analysed randomised or quasi-randomised controlled trials reporting absolute improvements in the percentage of patients receiving care recommended by clinical decision support systems. The findings were that in 108 studies reporting 122 trials with data from just over 1.2 million patients, clinical decision support systems increased the proportion of patients receiving desired care by 5.8% (95% confidence interval 4.0% to 7.6%). More importantly, in 30 trials reporting clinical endpoints, clinical decision support systems increased the proportion of patients achieving guideline based targets such as BP or lipid control) by a median of just 0.3% (interquartile range −0.7% to 1.9%), although to be fair studies where there was low baseline guidance adherence, and in paediatric settings, there were better results. The small to moderate improvements in targeted processes of care reported here should be cause for concern. This study can be viewed in full here.
Interestingly, another study in the BMJ has recently examined the clinical utility of treatment recommendations as these relate to age age and ethnicity of people in the United Kingdom, an observational cohort study conducted over a decade and assessing if adherence to clinical guidelines for hypertension translate to blood pressure reductions in current routine clinical care. The change in systolic blood pressure seen in new users of ACE Inhibitors/Angiotensin Receptor Blockers (ACE/ARBs) were compared to the results achieved with the versus Calcium Channel Blockers (CCBs), stratified by age (< v ≥55) and ethnicity (black v non-black), from baseline and at follow-up after 12, 26, and 52 weeks. A secondary analyses included comparisons of new users of CCB with those of thiazides. Essentially, this large study (which enrolled 87 440 new users of ACEI/ARB, 67 274 new users of CCB, and 22 040 new users of thiazides did not deliver results that were supportive of the guidelines. For non-black people without diabetes and who were younger than 55, CCB use was associated with a larger reduction in systolic blood pressure of 1.69 mm Hg (99% confidence interval −2.52 to −0.86) relative to ACEI/ARB use at 12 weeks, and a reduction of 0.40 mm Hg (−0.98 to 0.18) in those aged 55 and older. In subgroup analyses using six finer age categories of non-black people who did not have diabetes, CCB use versus ACEI/ARB use was associated with a larger reduction in systolic blood pressure only in people aged 75 and older. Among people who did not have diabetes, systolic blood pressure decreased more with CCB use than with ACEI/ARB use in black people (reduction difference 2.15 mm Hg (−6.17 to 1.87)); the corresponding reduction difference was 0.98 mm Hg (−1.49 to −0.47) in non-black people.In essence, the differential effects upon BP were not very large, and guideline adherence did not deliver superior results. And of course, the study did not and could not demonstrate that guideline adherence delivered superior results in terms of reduced adverse cardiovascular outcomes such as stroke or MI… See the study details here. An interesting perspective on the invidious effects of treatment guidelines, especially when in application driving polypharmacy in the elderly, has been published recently, and should be regarded as compulsory reading for anyone involved in clinical therapeutics in any way. The piece can be viewed here.
So here, published in the same journal, are research findings that challenge two different and central tenets of doctrine that is widely promoted and supported. So perhaps it would be reasonable to posit that Evidence Based Medicine (EBM) principles should be applied before advocating widespread conformity to guidelines. And that EBM principles should be applied before advocating for deployment of computerised clinical decision support systems to facilitate or even compel concordance with guidelines by using forcing functions? There are some other truths decision makers do not grasp, or choose not to see, in relation to the issues under consideration here. Questions that remain unanswered. These include:
- Is there really enough evidence to justify the allocation of PRODIGIOUS amounts of money to design, implement and maintain computerised clinical decision support systems?
- Do computerised clinical decision support systems (and their use to shape guideline concordance) actually deliver quantifiably superior outcomes (clinically, sociologically or even economically, if all costs associated with their use and effects are taken into account)?
- What do clinicians actually think about computerised clinical decision support systems? In essence, do they make life easier or more difficult, and at what others cost does the use of these systems come? Does compelling health workers such as doctors, pharmacists and nurses to use the systems actually compromise the safety, efficacy and efficacy of clinical care? It would appear that this question has never actually been thoughtfully posed, let alone definitively answered.
- If computerised clinical decision support systems do the “thinking” instead of people, how do future generations of clinicians learn to practice, especially in other settings where these sophisticated systems may not exist?
- If the same amount of money used to deploy and support non-human systems were to be allocated for the employment of human beings working to support safe and effective medication use, would the result be inferior, superior, or the same? Unless computers can be shown beyond doubt to produce superior results, why not employ people instead of spending money on machines? Remember, impressive financial and human return on investment has already been demonstrated with a simple approach deploying clinical pharmacists to guide safe prescribing and assist in training new generations of health care professionals who will practice in a variety of settings, including those not in tertiary referral centres (see details here). Of course there is a huge amount of money to be made in the clinical IT industry …
Computerised clinical decision support systems are often used to guide/push actions by clinicians in areas such as ordering tests or prescribing drugs. Rightly, many commentators call for attention to the influence of the pharmaceutical industry on the way in which drugs are prescribed. These commentators are often scathing in their assessments of “the pervasive and invidious influence of big Pharma,” a favorite villain of academics who make a living through critique. And to be fair, the pharmaceutical industry has a widespread history of culpability here. But interestingly, there is an apparent lack of scrutiny of the influence of “big IT.” Make no mistake, there are HUGE amounts of money to be made for any group that can successfully peddle a computerised clinical decision support system for deployment in a wide scale setting. It appears that the same critique for claims made about computerised clinical decision support systems, or even clinical guidelines, is not applied in a similar fashion to that expected in relation to drugs, where sponsor companies are compelled furnish extensive and expensive evidence to support their claims.
The rush to embrace the use of IT to direct clinical care, coupled with the widespread uptake of guidelines for clinical issues, probably has potential to enhance clinical care for patients. But it is unwise to take such things for granted, at least for now.