Redesign of a Transfusion Issue Form Using Human Factors Science

Transfusions at Children’s Hospital Colorado

During two observations of the hospital’s processes for performing blood transfusions, we identified multiple opportunities to improve documentation and assessments of patients. First, there was not a clear place to document vital signs (VS) in patient rooms during transfusions. The hospital does not have computers in all patient rooms; nor was there a standard form for documentation, which creates an opportunity for incorrect or incomplete documentation.

There was also confusion about when to assess core VS (heart rate, blood pressure, respiratory rate, and temperature) and when to complete additional assessments regarding transfusion hazards (examining patient’s skin, checking patient’s lungs, checking intravenous site). The varying time intervals and expected assessments at any given time period for transfusions added confusion for when to check different VS and additional assessments.

Our hospital uses a multipronged approach to guide practice for transfusion assessments and documentation, including annual training; a thorough policy; follow-up and feedback to nurses from transfusion medicine (Blood Bank); and a transfusion form. The transfusion form was intended to provide information for ensuring the unit was correct for that patient plus providing in-the-moment information and documentation for transfusion reactions and resulting actions. This form includes information on the correct unit, correct patient, VS, checkboxes and blank fields for symptoms of a transfusion reaction, and a list of actions to take following a transfusion reaction (Figure 1). However, the form was not as effective as it could be: the information was not in any apparent orders, the timing of VS was wordy and listed in a sentence format, there was no location to document VS when desired or during Electronic Health Record downtime, additional assessments were not listed, transfusion reaction symptoms were listen in a random order, and it lacked follow-up information for transfusion reactions needed by the Blood Bank.

OPEN IN VIEWER

Figure 1.

Original transfusion form.

The goal of this project was to review and redesign the transfusion form using human factors principles. We modified the transfusion form to a more informative and user-friendly design to provide a way for nursing to keep track of what VS and assessments were due, a mechanism to document those VS in the room, provide information in a usable format for identifying a transfusion reaction with steps for appropriate laboratory follow-up and documentation, and improve the flow of information use.

Human Factors Approach to form Evaluation and Redesign

Blood Bank management partnered with the hospital’s Human Factors Engineer to redesign the transfusion form to aid nursing practice. Throughout the redesign process, we applied multiple human factors methods to iteratively improve the form and evaluate effectiveness. During initial development, we performed a cognitive walkthrough with nurses and Blood Bank staff to identify key pieces of information and sequence of use. We asked the nurses to pretend as though they were completing a transfusion as part of these walkthroughs to improve sequence and display of text. Additional nurses and clinicians provided feedback on the original form during committee meetings. We also compared the form to policy to simplify the form – removing unnecessary text – and simplify the transfusion process by ensuring all necessary policy information was available on the form.

The transfusion redesign process incorporated multiple human factors design principles. These included Gestalt design principles (e.g., proximity, enclosure; Wagemans et al., 2012), visual hierarchy principles (e.g., order of information), other perception and cognition principles (e.g., text size, use of tables rather than text to amount of processing; Katz, 2012; Pettersson, 2010), and understanding of nursing workflow to influence alignment with processes (e.g., integration with context; Holden et al., 2013).

First, we performed an informal link analysis to show the order of information use (Figure 2). For even simple transfusions without reactions, nurses have to move around the transfusion form from the bottom to top. In transfusions with reactions, the order was worsened by going from the bottom to the top and then bouncing between the top and the middle sections. The order of information may have contributed to missed sections and items.

OPEN IN VIEWER

Figure 2.

Order of information used on original transfusion form. Blood bank performs steps 1-2 prior to sending the transfusion record to nursing. Green boxes and lines denote required steps for all transfusions, and blue denotes steps that are only required in cases of transfusion reactions.

We used the visual hierarchy of information typically going top to bottom for rearranging the order of content to align nursing sequence of use starting at step 3 (Nielson, 2006; Shrestha, Lenz, 2007). For the majority of transfusions where the patient does not have a reaction, we grouped all required information in the top section. We further reinforced this required section through use of Gestalt design principles of proximity and enclosure via a brightly colored border. We enclosed each of the main sections of required information, assessments, and suspected transfusion reactions to add clarity regarding the different sections of the form.

The previous process did not provide clear, usable information for when to obtain VS and perform assessments; VS and assessments could easily be missed. Information on VS and their timing was presented in sentence format, whereas additional assessments were not included on the transfusion form at all; policies with this information could only be viewed outside of the patient rooms that lacked computer access.

To reduce reliance on memory and improve visual comprehension, the text information on assessments was transformed to a table of time versus assessments with color coding. The table hierarchically organized information by which assessments were needed at a given time point, which aligned with mental models of nursing assessments at specific times, which reduced the level of processing. We used knowledge of human ability to identify images and patterns faster than written text to further improve the display of assessments needed at different times. Providing affordances via blank boxes versus dark gray boxes made it clear at a glance when to perform different assessments. This also provided a backup on the same form for documentation during downtimes and failures of the Electronic Health Record.

Transfusion reaction symptoms were present on the original form. However, there was no logical order to the listing of symptoms, and thus, symptoms were missed. Additionally, there was a lack of instructions or location to write follow-up information needed by the Blood Bank for specific symptoms (e.g., initial and ending temperatures for fevers), reducing efficiency while simultaneously creating gaps in processes. Therefore, we organized the symptoms first by frequency, whether additional information was needed, and then alphabetically. We provided affordances for documenting additional information by adding in labeled blank lines for providing additional information (e.g., “Ending temperature ___°C at time __:__”).

We also identified that the instructions following transfusion reaction symptoms were ineffective and underutilized. Anecdotally, many nurses did not recall that the information for steps to take following a reaction was available on the original form, while some information was only available in the policy. For example, the form did not clearly state that nurses needed to take the critical, time-sensitive step of immediately stopping the transfusion following recognition of any transfusion reaction symptoms. Although this step is provided in regular training for nursing staff, making this clear on the form is nonetheless likely to be helpful for nurses who infrequently transfuse. Information regarding necessary supplies to send to the blood bank was contained in a policy that had to be accessed electronically, separate from the paper transfusion form. Additionally, contact information for the blood bank was located in a third location. Therefore, we created a high-level section of suspected transfusion reactions with sublevels of symptoms and actions needed. We created a table of supplies by product type. We used color, bolding, and capitalization to draw attention to the critical, time-sensitive step of stopping the transfusion. We also added contact information, highlighted for salience. We improved readability of this section by increasing font size and line spacing.

Ultimately, we updated the sequence, grouping, contents, and display of information of the transfusion form (Figure 3). The sequence on the transfusion form better aligned with nursing use (Figure 4). We provided clear headings and separate enclosed sections of information. We moved information from Knowledge-in-Head to Knowledge-in-World by visually displaying interactions and providing additional information from policy. We also increased readability by increasing font size throughout the form; however, this necessitated use of a legal document rather than a standard letter size for several months. Text was shortened to reduce noise. These design interventions resulted in the updated transfusion form.

OPEN IN VIEWER

Figure 3.

Redesigned transfusion form.

OPEN IN VIEWER

Figure 4.

Order of information used on redesigned transfusion form. Blood bank performs steps 1-2 prior to sending the transfusion record to nursing. Green boxes and lines denote required steps for all transfusions, blue denotes steps that are only required in cases of transfusion reactions, and the purple step is optional.

Usability Testing

We performed usability testing with eight nurses from hospital units and services with frequent transfusions (apheresis, ECMO [extracorporeal membrane oxygenation], Center for Cancer and Blood Disorders – inpatient and outpatient, neonatal intensive care unit, and pediatric intensive care unit), with the goal of testing at least one nurse from each of those areas and at least five nurses in total (Turner et al., 2006). Nurses used the original and the redesigned forms with mock patient information for scenarios with and without transfusion reactions. We also developed a knowledge assessment on when to perform different assessments and identification of symptoms that was administered pre- and posttesting, not presented in this article. Following all scenarios for each form, nurses filled out a perceptions questionnaire that contained items such as perceived usefulness and perceived ease of use. Finally, structured interviews were performed to identify barriers, facilitators, and resources for the use of the different forms and the preferred layout.

For the original transfusion form, the most commonly identified aspect of success was existing knowledge and experience using the form (“I know how to use the form”). Nurses had developed their own tools for improving task completion, such as creating self-made checkboxes on a separate assignment sheet (“This is the way I’ve always done this”); updates to the transfusion form provided affordances to reduce these work-arounds and excess effort. However, they did note several issues with the original form. Documentation could be challenging because there was no room on the form to write down assessments, and their existing strategies did not help organize documentation in the same place. Assessments were challenging due to variations in timing; a planned mitigation in the Electronic Health Record was not specific enough to help. Additionally, nurses reflected that they sometimes missed documenting final assessments following transfusions – an example of a postcompletion error. Finally, although policy and procedure information were present on the form, nurses recognized that it was so cluttered that they stopped paying attention to noncritical blocks of text, further demonstrating the need for a more usable form.

Overall, five of the eight nurses who participated in usability testing preferred the redesigned form (Figure 5). Of note – they mostly believed that the redesigned form would be most helpful for new graduates (“I already know what I’m doing”), but that the algorithmic nature would benefit all nurses. The changes moving information from Knowledge-in-Head to Knowledge-in-World clarified requirements for each step in the transfusion process (“My brain doesn’t have to think”). The nurses liked that there was an optional location for documentation that they thought would be especially helpful in certain situations (e.g., no computers available). Finally, nurses liked changes to the formatting – text size (especially for reading in low-light situations, “Aging eyes”), reduction of text, removing blocks of text, and grouping of sections with separation between sections.

OPEN IN VIEWER

Figure 5.

Usability testing (N = 8): Counts of nurse preferences for the original (“current”) versus redesigned (“proposed”) form.

However, the redesigned form was “overwhelming” and “seems like extra work at first glance.” Nurses recognized this would require change to documentation practices. Although many nurses liked the optional assessment documentation table, one nurse thought this unnecessary (“No different from writing on a paper towel”). Additionally, the longer paper length was universally disliked – it did not fit well with other charting and folders. Only one nurse preferred the original form (“I don’t like change”).

Postimplementation Assessment

We then sent a follow-up survey assessment to the nursing staff after 7 months of using the redesigned form. We had 67 registered nurses respond to the survey who had given blood using this new form. In this survey, we asked for specific evaluations of each form, including likes and dislikes for each of the main sections: required information, assessments, and suspected transfusion reactions; UMUX (Usability Metric for User Experience)-Lite questions to provide a system usability equivalency score (scale of 22.9–87.9; Lewis et al., 2013); and overall form preference. The project was classified as a Quality/Improvement project (QI #1807-5, QI #1904-4) rather than research by the Organizational Research Risk & QI Review Panel at Children’s Hospital Colorado.

After 7 months of use, approximately half of the 67 nurses who completed the postimplementation survey preferred the redesigned form (Figure 6). Similar to usability testing, the main facilitators for the redesigned form were the display and organization of information, ease of use, and useful sections and information. The redesigned form had greater perceived usability than the original form with a medium effect size (t = 3.14, p < 0.005; Figure 7).

OPEN IN VIEWER

Figure 6.

Survey (N = 67): Counts of nurse preferences for the original (“original”) versus redesigned (“current”) form.

OPEN IN VIEWER

Figure 7.

Survey (N = 67): UMUX-Lite scores for the original (“original”) versus redesigned (“current”) form.

Although the majority of nurses liked the redesigned form, it was not universally preferred – approximately one quarter of nurses preferred the original form. Analysis by department showed trends that hospital units that transfuse blood infrequently (e.g., medical–surgical floors) tended to be more positive about the redesigned record than units that transfuse frequently (e.g., Center for Cancer and Blood Disorders, perioperative areas). These trends persisted in both the usability data and the overall form preference.

The main barriers to use of the redesigned form mirrored that of usability testing: too much information present, disliking change, and personal preferences – mostly related to the first two points when specified. An additional emergent barrier was the high variation in use of sections; many nurses chose not to use the assessments section. In the likely case that the nurse chose not to use the assessments section and the patient did not have a suspected transfusion reaction, the majority of the form was not used and, thus, wasted space.

Conclusions

Forms are ubiquitous across industries and contexts. Although forms can be viewed as simple and insignificant, their designs can greatly affect workflows and outcomes. A form is an example of a cognitive aid to help people complete tasks. In this case, the transfusion form was intended to help nurses with assessments and documentation for transfusions, which has the potential to affect patient safety. However, the form as designed originally was challenging to use.

Our redesign work resulted in a streamlined transfusion form that improved readability, comprehension, knowledge of necessary follow-up action, and alignment with nursing workflows. In addition to being more usable, the redesigned form was also generally preferred. Documentation of assessments and symptoms of transfusion reactions have increased while calls to the Blood Bank for clarification on process questions have decreased per the anecdotal observations of the Medical Director of the Blood Bank. These outcomes show the value of the redesigned form.

However, many end users still preferred the original record due to dislike of change – a barrier that may not be overcome by design. To this end, our multipronged redesign project shows the value in capturing user’s perceptions and qualitative experiences, rather than focusing solely on standard metrics such as errors or time to complete. Future work could include incorporating additional change management techniques to improve acceptance.

We identified multiple issues and barriers to use for both the original form and the redesigned form. For example, copying information from the policy onto the form provides information at the time it is needed; however, this increases risks of the form not being updated alongside relevant polices. Although we provide education and training to all nurses on how to effectively and safely perform and document blood transfusions, this project demonstrates that these mitigations are not always effective in improving performance. Work-arounds exist and commonly point to gaps in system designs. While creating a usable form is a more robust mitigation than education, this cognitive job aid will not fully prevent errors.

There were multiple limitations to this work. Notably, we did not have project resources to collect thorough baseline and postintervention data, such as long-term acceptance of the redesigned transfusion form. Additionally, we did not collect data beyond preference measures to compare differences in outcomes such as completeness of documentation, nursing knowledge of assessment timing, efficiency (e.g., time, eye tracking movements), errors, or patient outcomes – we acknowledge that these data would be useful in understanding the full impact of the form redesign.

The transfusion form project provides an example of how to incorporate human factors principles and methodologies in a context (blood banking) that historically has not worked with human factors professionals. Specifically, we incorporated principles of Gestalt design principles, visual hierarchy principles, other principles from perception and cognition, and a thorough understanding of nursing workflow to redesign the form. The redesign process was followed by feedback from end users, usability testing, and a survey to evaluate the redesigned form. These design principles and methods are transferable to other forms and processes for improvement. Ultimately, we encourage other professions to consider the interactions of people, tasks, technology, and contexts when designing or evaluating processes – including interventions as “simple” as a form.