Developing the new tool. The HPR was developed by 3 wound care experts — 1 nurse and 2 physical therapists — each with more than 15 years of experience in wound assessment) using the PUSH tool as a template. Over the course of ~1 year, 1 of the experts began to outline the HPR and to casually experiment with it in practice to see if different items and scoring structures could indicate numerically whether wounds were the same, better, or worse compared to expert opinion. The use of the PUSH score at the expert’s hospital facilitated comparison with HPR scores. These efforts were fueled by wound care team members at the hospital expressing frustration with the limited tissue type choices in the PUSH and the fact that PUSH scores showed no change in larger wounds that were healing. For example, the PUSH tool has no option for purple as the tissue type (as is seen in DTPI), so clinicians would have to decide to call it black (a score of 4) or red (a score of 1). Thus, a purple wound bed could go from a score of 4 on admission to a score of 1 on discharge without any change in wound status.
Table 1 presents the new tool. It comprises 3 sections: worst tissue type observed, surface area, and whether skin is intact. Items in each section are assigned a point value and tallied for a total score. Based on the casual experimentation done during clinical care, the same scoring and score weights used in the PUSH tool were used and expanded to include a larger surface area scale and a larger, more detailed tissue type scale. The other 2 experts reviewed item order and point values for face validity. The PUSH tool exudate section and its subscore were eliminated in the HPR because of the expert’s concern over the subjectivity of this metric. The 3 experts all believed exudate was important but could not be objectively verified in photo-documentation when nonexpert clinicians performed assessments.
Tissue types that have been identified by the NPUAP as essential markers for PI stages were added and include maroon or purple, blood- or serum-filled blister, and red granulation tissue separate from smooth red dermal tissue.24 Additional size categories were added to raise the ceiling on surface area to 225 cm2. Scores for the new tissue types and size categories were ordered based on NPUAP stage association or size and given the next whole number as their respective value; for example, a serum-filled blister would be scored 2 while a blood-filled blister would be scored 5. A new objective subscore for surface skin layers (intact or not intact) was added to represent a healed wound, a DTPI, Stage 1, intact blister, or intact eschar; this intact skin score was given a value of 0.5. The tissue type and size scores are in whole numbers and together could range from 0 (best — ie, less severe wound) to 27 (worst, most severe wound). The 0.5 addition for intact skin would clearly communicate the presence of an open wound versus a wound with intact surface layers because the tissue type and size scores are in whole numbers.
Once the new tool was developed, Institutional Review Board approval was obtained from the hospital where data collection would occur as well as from the university where data would be analyzed.
Photographs. The hospital obtained consent from all patients for deidentified pictures to be used in research. Photographs from hospital medical records were utilized and did not include any personally identifiable patient information. The pictures were selected by the research team to represent a variety of PIs and patient types to compare PUSH and HPR scores for different wounds. This was a nonrandom sample of wound images without specific inclusion or exclusion criteria beyond the requirement that the pictures be of the same PI taken at 2 clearly identified points in time. If the wounds were not clearly PIs with date and time marked, they were not considered for inclusion. Photographs of the wounds were needed at 2 different time points to compare wound change for each wound using both PUSH and HPR. With these goals in mind, 30 cases were selected and 2 photographs of the same wound were selected from 2 different points in time. Wound types and locations are shown in Table 2.
All wound photographs at this hospital are taken with the NE1 Wound Assessment Tool (NE1 WAT, originally called the NE One Can Stage “NEOCS”, N.E. Solutionz, Las Vegas, NV) framing the wound.32,33 The NE1 WAT is an “L” shaped ruler with a slightly adhesive backing designed to aid in the assessment of PI. The tool has representative pictures of PI for the various NPUAP stages with descriptions of those stages. It also contains room for documentation of the patient, provider, date, and time of assessment. In 2011, the development and psychometric testing of the NE1 WAT was described and its reliability and validity were established.32 In the first of these tool psychometric studies, 101 clinical wound care providers (registered nurses, physical therapists, physicians, and nursing and physical therapy students) staged pictures of wounds with and without the NE1 WAT.32 In the second, 94 registered nurses staged pictures of wounds with and without the NE1 WAT.33 In both studies, reliability coefficients were high (ICCs = 0.6–0.9) and accuracy of clinician wound staging improved by as much as 61% when they used the NE1 WAT compared to when they did not.32,33 However, it does not provide a score indicative of wound healing or worsening.
Scoring by experts. Using print photographs and paper-and-pencil instruments, the 3 expert participants that initially developed the HPR viewed each of the 60 images (30 cases) and scored the wounds using the PUSH and HPR tools, yielding 3 scores for each case. Because the measurement instructions are different between the tools, experts were cautioned to strictly follow the directions of each tool for each element of scoring. For example, the PUSH tool requires measurements to be taken from the wound edge and the HPR directs that involved periwound should be included in the size measurement. Additionally, the instructions for obtaining length and width measurements are worded slightly differently. The HPR directs that measurements for length should be from 12 o’clock to 6 o’clock and width from 3 o’clock to 9 o’clock at the widest and longest portion of the wound, and the PUSH tool measurements are obtained using greatest length (head to toe) and the greatest width (side to side). The wound images were all taken with the NE1 WAT on the patient framing the wound, and the ruled border of the NE1 WAT was used for measurement in this assessment. These 3 experts then met to discuss their scores and came to consensus so each picture had 1 score from PUSH and 1 score from HPR. The HPR and PUSH scores for each case then were compared to determine if they were lower, higher, or the same in terms of change between the 2 photographs.
A convenience sample of 6 additional wound care experts (4 nurses and 2 physical therapists with a minimum of 10 years of clinical wound experience) were recruited via email. When the invitation was accepted, the criteria and further instructions were emailed to them and they were asked to review the same cases and existing expert scores and provide feedback comparing the HPR to PUSH along with suggestions on how to improve the HPR. In cases where the PUSH and HPR scores disagreed regarding direction of wound healing (same, better, worse), the 6 experts were asked to indicate if they thought the wound had improved, deteriorated, or remained the same. Once the test packets were sent out, a time frame of 2 weeks was given for completion.
Nonexpert testing. A convenience sample of 120 participants was recruited from a university and a community hospital. Participants were approached in person by members of the research team in the course of their work as students or clinicians in the university or hospital over a period of 2 months. If persons recruited met the inclusion criteria (ie, educated on PI management and assessment and expected to perform these skills in their professional roles now or in the future), they provided written informed consent and completed a demographic profile on paper and were assigned a subject number. The demographic profile included years of experience in wound care, specialty certifications related to wound care, a single question on self-perceived skill level in wound care (Likert-style scale with 5 choices from expert to none), and prior experience with either of the tools being tested (yes or no) (see Figure 1 and Table 3).
The participants used a computer in a hospital training room to view 30 cases (60 images) and score the photographs according to the HPR and the PUSH tools. The computer program was a training version of the clinical documentation interface used by the hospital for wound care documentation. Screenshots of the interface are provided in Figures 2, 3, and 4. Participants were asked to read the scoring directions for the PUSH tool and then use the tool to score a semi-random selection of 10 of the 60 pictures. The study was designed so that each picture would be shown to at least 3 participants. Participants then were asked to read the scoring directions for the HPR and score a different semi-random selection of 10 pictures using the HPR. No other training was provided during the study. The computer program used a digital box, the size and placement of which was controlled by the participant, to measure wound size. Scoring was completed automatically by the computer program based on the participants’ selections.
Following the computer-based wound scoring exercise, all participants completed an exit survey on paper (see Figure 1) to provide feedback on the 2 tools that included 3 questions with Likert-scale responses and 1 free-response item. The first 2 questions asked the participants about their preference for HPR or PUSH and the Likert scales had “0 1 2 3 2 1 0” options with PUSH on the left and HPR on the right. The third question had a traditional 7-point Likert scale with 1 = not important and 7 = very important. All testing and survey completion was completed in a single session which lasted approximately 1 hour for each participant. The testing room had 10 computers and groups of participants were tested simultaneously whenever possible without allowing them to discuss the cases or their answers during testing.
Data collection and analysis. Nonexpert scores for each item of the PUSH and HPR were exported directly into a spreadsheet file; data from paper forms were manually entered into the same spreadsheet. Following data collection spreadsheets were imported and analyzed in SPSS Version 24 (SPSS Statistics for Windows, IBM Corp, Armonk, NY).
Descriptive statistics were generated and a paired t-test was performed to compare the mean score (HPR and PUSH) of nonexpert participants on each image to the expert score for each image. Images then were placed into 3 groups based on the result of the t-tests. If the mean of the participants’ score was significantly higher or lower than that of the expert for an image, that image was placed into the respective (“higher” or “lower”) group. If no statistical difference was noted between the participant mean and expert score, that image was placed in the “same” group. Chi-squared tests (χ2) were used to determine whether the frequency distribution within each of the above-mentioned 3 groups was significantly different between the 2 types of tools used (HPR and PUSH). Due to violations of normality, the nonparametric Friedman’s analysis of variance (ANOVA) and Wilcoxon signed rank test were used to compare the experts’ votes on which tool they thought captured the difference between the pre- and post-wound picture and to perform post hoc testing, respectively. Free-response comments to question 4 of the exit survey were grouped into common themes by the authors in a group setting.