- Evidence suggests health IT-supported clinical pathways can improve patient outcomes, but more and better studies are needed to demonstrate the true effect, a review in the Journal of the American Medical Informatics Association finds.
- The U.K. researchers reviewed 44 studies involving more than 270,000 patients. Of those, 86.4% cited benefits associated with HIT-supported pathways, including quality of care (65.9%), improvements in objectively measured patient outcomes (34.1%) and healthcare resource utilization (22.7%).
- Clinical decision support was the most frequently reported HIT intervention, used in 56.8% of the studies, followed by modified electronic documentation and computerized provider order entry, at 52.3% each.
As more digital health technologies become available, providers are looking for clinical evidence that tools like machine learning and electronic order entry do what their developers claim and actually improve patient outcomes.
Some early studies point to the potential for HIT to increase efficiencies among multidisciplinary teams and inform diagnosis and treatment decisions. An analysis last year in Nature Medicine found that Google's DeepMind artificial intelligence system could detect and recommend how patients should be referred for more than 50 eye diseases with the same accuracy as a physician.
While early, the results suggest DeepMind not only is capable of handling the spectrum of patients seen in routine clinical practice but also speeds analysis of scans — handling more than 1,000 a day, the DeepMind team blogged at the time.
Meanwhile, a 2017 report found the U.S. could save $7 billion in health spending annually by broadly deploying digital health apps for high-cost patient populations such as those with diabetes, asthma and cardiac rehabilitation.
The authors of the JAMIA review, apparently the first to systematically assess the effects of implementing HIT-supported clinical pathways, found that while most studies hinted at benefits when digital tools were employed, the lack of homogeneous study designs and risk of bias make it hard to draw conclusions about the impact of deploying such pathways.
For example, five studies saw effects of HIT-supported pathway interventions on patient outcomes such as lower rate of central line-associated bloodstream infections and lower hazard ratios for heart attack and stroke. But the heterogeneity of outcome measures made it impossible to conduct a meta-analysis.
Overwhelmingly, the studies also focused on the benefits of HIT interventions and ignored possible disadvantages or risks. Of the 44 studies reviewed, just three included adverse events associated with the use of HIT-supported clinical pathways. Most included no methodologies for identifying or reporting unfavorable findings.
"Given the ongoing investment in HITs there is a need for investigation of the implementation of these technologies using robust study designs that minimize the risk of bias," the authors write.
"It would also be helpful for future evaluations to carefully consider the outcome measures that they report, to ensure that these are likely to be of relevance to key stakeholders," they add. "Ideally these outcome sets should include objectively measured patient outcomes and adverse events to allow a full appraisal of the effects of the intervention being investigated."