Preventing provider errors: online total parental nutrition calculator

From Clinfowiki
Jump to: navigation, search

Preventing Provider Errors: Online Total Parenteral Nutrition Calculator

Christoph U. Lehmann, Kim G. Conner, Jeanne M. Cox

Introduction

The Institute of Medicine report in 1999 indicated that more than a million injuries might be caused by medical errors annually as a result of manifested Adverse Drug Events (ADE’s) mostly in the medication processing work flow, from prescription to the point of administration. The rate of potential errors to occur during clinical medication intervention is three times higher in children than adults, it could be higher for the premature newborns at the neonatal intensive cares units (NICU’s), not only for medication, but also includes errors in parenteral nutrition ordering.

An effort by a group of researchers at the NICU of the Johns Hopkins Hospital was aimed to prove that practical interventions to change system processes are more successful than changing people in preventing ADE’s. After two years of designing, utilizing, and improving a low-cost online nutrition order entry system (TPNCalculator) instead of paper-based system that was identified as the main source of errors and been in use for over 10 years. They were successful in reducing errors in parenteral nutrition ordering and calculating by 100%.

Information to evaluate the effectiveness of the TPNCalculator was captured during three different phases, each lasting for about a month and half, a control phase, which was before using the calculator and was based on paper-form ordering. Phase two immediately after the implementation, while phase three started little bit over two years from the first phase. During phase one, NICU average patient census was 32.3, phases two and three were comparable at 30.2 and 32.4. During phase one, Total Parenteral Nutrition (TPN) orders were 557 with an average per patient per day of 0.39 and errors rate of 10.8 per 100 TPN orders (that required the pharmacist to contact the provider).

During phases two and three, the TPN orders were 471 and 656, the errors rate per 100 orders dropped to 4.2 and 1.2 respectively. Phase two showed 61% reduction in errors rate and significant reduction in problems, including 100% reduction in calculation errors, 88% reduction in osmolality guidelines misuse, and 84% reduction in other knowledge problems. One issue reported during phase two was a 35% increase in the number of incomplete forms (orders without order identification -page number) which was corrected during phase three and totally eliminated.

Phase three, continued with the same success level of phase two, producing 89% reduction in errors rate and slightly better reduction in problems, including 100% reduction in calculation errors, 91% reduction in osmolality guidelines misuse, and 100% reduction in other knowledge problems.

As the user (providers, nurses, and pharmacists) accesses the TPNCalculator via an Internet browser from a public workstation or a desktop computer (using individual log-in identification and password) to provide his/her input, add new patient, modify/renew an old order for an existing patient, the TPNCalculator performs all necessary fluid and component calculations. Based on nutritional guidelines, an osmolality calculator, and 62 rule-based alerts and reminders, the TPNCalculator creates alerts and reminders including a check for age/weight ratio, dose range alerts based on total dose, dose per weight, and concentration, it also suggests nutrition based percentage of total calories. Other features include automatic selection of the appropriate protein additive based on age and an osmolality alert. The TPNCalculator provides a dynamic calculation, validation against a possible range, and reminders based on previous user selections and on going changes of orders, prevents order submission unless it contains patient name and medical record number, physician name, and identification number, and automatically populate hospital admission, discharge and transfer systems with patient data.

As a quality improvement instrument build form a routine web development tool (Macromedia Cold Fusion) and designed to resemble closely the TPN order form, it required minimum user training. The TPNCalculator development and testing (which took three weeks by the participant/observers) was based on focusing on the underlying reasons of an already-existing process failure, this method protected the existing order processing flow from any possible changes.

Assessing its acceptance by the ordering providers, the TPNCalculator was compared with the paper based ordering form, users indicated via questionnaires that it was found to be easier to learn and to use, protects against errors, saves time, helpful for data entry, does not cause data overload, and constituted an improvement.

Comments

As HIS systems development guidelines suggest many rules to ensure accuracy, reliability, and consistent intended performance that the development of this application did not totally follow, yet, the application was a success due to the following factors:


  • The software was modeled on an existing paper form in use, which implicitly determined system inputs and outputs, functionality, interface definition, user interactions, safe ranges, limits, and default values.
  • Using domain experts for the testing allowed for testing of all potential clinical scenarios using patient data available during the development period.
  • The robustness of the system was tested in a clinical environment by allowing users to use TPNCalculator while for a period of 3 months all calculations were duplicated in the pharmacy to ensure safety. During that period, users generated over 1800 orders, which spanned the complete range of possible inputs.
  • The decision by the management to provide an Internet browser on all workstations was crucial to implementation.
  • The risk was minimized by using participant/ observers in the development

Lessons to learn from this tool implementation are not limited to the fact that rapid cycle evaluation is able to handle software design problems through frequent series of design, evaluation, and redesign. Furthermore, includes the fact that using existing infrastructure to develop a pragmatic medical informatics solution to the problem while leaving the remaining system intact is possible and proven to be successful in both reducing the number of errors and increasing user satisfaction.

References

  1. http://pediatrics.aappublications.org/content/113/4/748.short