Dear Dr Sanford, Editor,

Please find this correspondence in response to the article published by Nursing Administration Quarterly, titled "Bayesian Cost-effectiveness Analysis of Falls Risk Assessment Tools," by McNair and Simpson. There are several concerns raised after reading this publication. From a methodological perspective, their first error was to apply an economic analysis to 2 programs that are not equal in scope, design, breadth, and IT complexity. In addition, the MFS is a screening tool, not an assessment tool. The authors also utilized findings of the reported sensitivity of the HDS in a neuro population and generalized those findings to a general population of patients. A more accurate representation of the sensitivity and specificity of the HDS for general populations was presented at the American Medical Informatics Association annual conference in 2014. The HDS was revalidated after being translated for use in the electronic health record (EHR). Sensitivity and specificity were reported at 90.3% and 64.8%, respectively. Given the authors' affiliation with Cerner Corporation, they should have access to these publications.

In addition, the authors mischaracterize the value of specificity over sensitivity. When predictive analytics are utilized in care, it is rare that we can make a prediction and let fate take its course. Instead, we must aggressively work to prevent what it is we have predicted. This can and does negatively impact specificity where those cases that did not experience the outcome predicted would have except the active efforts of caregivers prevented it. This treatment paradox is well described in our publication cited by the authors but is not mentioned in this publication. This is a critical concept for nurse executives and scientists to understand and account for when making decisions. Sensitivity is the more important measure of performance, as it reflects that ability of the tool to accurately predict positive cases. In the matter of falls, the more cases you miss, the more you can expect to see poorer patient outcomes related to fall and injury rates, as well as higher risk for litigation that cannot be adequately defended.

Another concern is the overall design of the study that evaluates 2 fall prediction tools without accounting for the entire nursing process. These tools are only the foundation of a larger process of care. The HDS, for example, is intended to be used as a fall risk assessment tool that identifies which patients are at risk, the level of that risk, and why the patient is at risk. When coupled with the HD Falls Care Plan, it is a powerful approach to care that has proven time and again across patient populations and organizational levels to reduce both falls and falls with injury. It is because the HDS assists the clinician in using the HD Falls Care Plan to efficiently drive the right intervention to the right patient at the right time. This program has reduced event-related costs and also eliminates the need for sitters/observers for fall management. At UAMS Medical Center, when the HD Falls Program was deployed in 2011, the estimated cost savings related to falls and injury in a single year was $1.2 million. The elimination of sitters continues to save another $330 000 annually in unnecessary labor costs. It is the plan of care that does the work of driving outcomes. Prediction is only the start. To be fair in a cost analysis, the entire nursing process needs to be accounted for.

It is appreciated that the overarching message to nurse executives is that when making decisions about content and changes within the EHR, there needs to be consideration for cost. What is missing here is that there are multiple financial angles to be considered including cost to change and cost to implement, as well as cost aversion including the elimination of unnecessary care, reduction of care for hospital-acquired conditions, and avoiding penalties and litigation. Cost-effectiveness is often a cycle, one which is not fully represented here. The true measure of success is improved health and function. This analysis treats falls as a binomial outcome: fall prone/not fall prone, but fails to cost out mitigation or elimination of actual fall risk factors. The true measures of fall risk prevention programs is to reduce overall risk and harm events both of which are supported by the HD Falls Program.

-Amy L. Hester, PhD, RN, BC

Director of Nursing Research and

Innovation

UAMS Medical Center

Little Rock, Arkansas

Chief Scientific Officer

HD Nursing, LLC Benton, Arkansas

[email protected]

-Patricia A. Quigley, PhD, ARNP,

CRRN, FAAN, FAANP

Nurse Consultant

[email protected]