Warning States for each patient are calculated by RI Server and are sent to RI Mobile. RI Server allows for the configuration of rules which support the preferred warning state logic of the organization. Currently, RI Server offers a three-tiered approached to Warning states which are Very High, High, and Medium.
Warning states are based upon “rules” (criteria) related to the RI score. For example, if a patient’s RI score has dropped by a predefined amount or has decreased below a pre-defined threshold.
Additionally, different patient populations (e.g. adult, pediatric, ICU or non-ICU) may be assigned different rules. The choice of criteria and mapping for each warning state level is configured during the RI Mobile implementation process.
The patient lists categorize patient graphs by current warning state. For example, patients meeting the rule criteria of the “Very High” warning state will be categorized with a “Very High Risk” warning designation.
It is important to note that it is the value of the RI scores that relates to the likelihood of an adverse event, not the change (i.e. drop) in the RI score. The validation of the RI has demonstrated that lower valued scores correspond to increased risk. The purpose of configuring a warning state that is triggered by a change in RI is to help direct attention to patients for whom a lower RI score is a new situation. Particularly in instances where the low score is a recent development, a change-based warning can help ensure clinician awareness in the event that a clinician’s last observation of a patient’s RI score was when it was at an earlier, higher value.
Warning State Hierarchy
During an active warning state, a higher-level warning can override an existing, lower-level warning. A higher warning state always overrides a lower warning state:
- Medium
- Overridden by High and Very High Warning States
- High
- Overrides Medium Warning State
- Overridden by Very High Warning States
- Very High
- Overrides Medium and High Warning States
A lower level warning state will not override an existing higher-level warning state; however, a lower level warning state may take effect following a higher-level warning state elapsing (i.e. expiring) if the last data point before the warning state elapse time meets lower level warning state criteria.
Warning State Logic
It is important to note that timely warning states are dependent on timely nursing documentation. For example, if a nurse enters data at 1700 for a nursing assessment that occurred at 0800 and these clinical values generate an RI score that leads to a warning state, the RI Time will display at 0800 even though the warning state will not have been visible to users until the time of the documentation entry at 1700. As a default configuration, warning states are active for 24 hours from the timestamp of the triggering RI score.
Warning State Selection
Warning states are configured and used by customers to support their organizational initiatives and goals. The clinical protocols or actions associated with warning states, such as reviewing a patient’s record or re-assessing the patient, are determined by the organization. Depending on an organization’s goals and resources, it may be appropriate to have a larger or smaller proportion of patients in warning RI Mobile warning states. Spacelabs provides product implementation support to assist customers in determining how they wish to configure warning rules. This support includes:
- Assistance in determining implementation approach according to the customer's organizational goals, resource constraints, and operational preferences. This may also include reference or referral to other customers to share learning and product experience.
- Analysis of the customers historical data to estimate how a given set of warning rules will perform with the customer organization's own patient population.
- Review of documentation delays (i.e. delays between the time of clinical measurement and the time of data input) to evaluate impact on the utility of warnings.
The analysis involves a retrospective review of customer data (usually 12 months of inpatient data) to evaluate how various warning rules are expected to perform. An evaluation of data entry delays is included to assess the impact that such delays may have on warning performance. Warning performance metrics typically include:
- The proportion of patients overall who are flagged by a warning during their admission
- Indicates volume of warnings.
- The proportion of patients who die in the hospital and are flagged by a warning prior to death.
- Indicated how sensitive warnings are to the most critically ill patients.
- The numbers of new warnings, on average, that are likely to be generated during each 12-hour nursing shift.
- Indicates day-to-day volume of warnings which may impact clinician burden and an organization's expectations for response.
Such analysis will assist an organization in determining the sensitivity and frequency of warnings most appropriate to the organization.