Harriet Lane Online Culture Book: Matching Residents' Workflow

Abstract

Managing cultures taken from patients in a teaching outpatient setting, such as Johns Hopkins Harriet Lane Clinic (HLC) in Baltimore, is a complicated task to model, because it must support a task that lasts several days. There are many considerations in replacing the paper based system currently in place. We needed to balance the need for speed with the recognizability of familiar forms; the need for quality care with the need for teaching; and the need for documentation with the need for compact representation of clinical need.

The process of refining the scope of the software system was iterated through many users and stakeholders at the HLC and other pediatric outpatient venues (Emergency Department and Adolescent Clinic). Revisions were made to distill the critical components and determine the most effective interfaces.

One frequently cited claim for the reluctance to switch to a digital recording system is the time necessary for data entry. We have devised several methods to reduce the burden of paperwork and aid in the rapid entry of frequently used cultures. On the one hand, we offer a rapid interface for the ten cultures that account for 80% of all HLC cultures. On the other hand, we recreate the functionality of current paper order forms through a series of hierarchical menus that offers every permutation of every possible sets of orders available at the HLC. We estimate that the pick list entry screen will capture 95% of all ordered cultures, however evaluations need to be done to determine actual efficacy.

Design-component needs required reconsideration of conventional methods of data selection. For instance, a concern voiced by attending physicians was the desire to limit implicitly advising users in the selection of medications, so they would reason out the choices. This concern precluded the use of pick lists, which may lead to higher quality care. So, a text box was used with additional text boxes for dosing, delivery and frequency. Units, delivery and frequency were defaulted to the most commonly used types. Future revisions may dynamically adjust default values based on available data. Immediately this method of input introduced the problem of data integrity to a normalized database. This problem was addressed using the Soundex algorithm to interpret the drug name. If the entered data matches the lexical definition in our drug database then the input is simply accepted. Otherwise, if the drug name is incorrectly spelled or more than one drug resolves to the same Soundex then the user is asked to verify the correct information before proceeding.

Workflow model of software culture book for Harriet Lane Clinic

Documentation involved the actions taken each day as new data from the microbiology lab became available. An entirely new process was initiated (see Workflow model) that fit into a single Web-based screen.

Conclusions

The process of replacing an established method of clinical documentation must be perceived of as better than the method it is replacing, for instance, through the reduction of paperwork and redundant documentation. The standards for computer based systems are much higher those of traditional recordkeeping. In our experience a simple concept quickly becomes a major undertaking involving many stakeholders. It is best to consult as many stakeholders as possible and proceed accordingly. Our iterative model of software design resulted in a more cohesive and clinically relevant documentation system. The design is also extensible for study of users’ behaviors.