Adaptation and early implementation of the PREdiction model for gene mutations (PREMM5™) for lynch syndrome risk assessment in a diverse population

Abstract

Lynch syndrome (LS) is the most common inherited cause of colorectal and endometrial cancers. Identifying individuals at risk for LS without personal cancer history requires detailed collection and assessment of family health history. However, barriers exist to family health history collection, especially in historically underserved populations. To improve LS risk assessment in historically underserved populations, we adapted the provider-facing PREdiction Model for gene Mutations (PREMM₅™ model), a validated LS risk assessment model, into a patient-facing electronic application through an iterative development process involving expert and patient stakeholders. We report on preliminary findings based on the first 500 individuals exposed to the adapted application in a primary care population enriched for low-literacy and low-resource patients. Major adaptations to the PREMM₅™ provider module included reduction in reading level, addition of interactive literacy aids, incorporation of family history assessment for both maternal and paternal sides of the family, and inclusion of questions about individual relatives or small groups of relatives to reduce cognitive burden. In the first 500 individuals, 90% completed the PREMM₅™ independently; of those, 94% did so in 5 min or less (ranged from 0.2 to 48.8 min). The patient-facing application was able to accurately classify 84% of patients as having clinically significant or not clinically significant LS risk. Our preliminary results suggest that in this diverse study population, most participants were able to rapidly, accurately, and independently complete an interactive application collecting family health history assessment that accurately assessed for Lynch syndrome risk.

Keywords: Family history; Hereditary cancer; Lynch syndrome; Risk assessment; Underserved.

Fig. 1

Depiction of iterative adaptation and implementation of the patient-facing version of PREMM₅™. Following initial adaptation of the provider-facing tool, a health literacy expert performed literacy adaptation. The English-speaking PAC provided feedback on these materials, and this feedback was incorporated into wireframes that were iteratively reviewed by the study team and English-speaking PAC. The Spanish-speaking PAC provided iterative feedback on crucial aspects of the culturally-coherent translation

Fig. 2

Patient pathing through study eligibility criteria assessment, including the PREMM₅™ tool. Eligibility by PREMM₅™ corresponds to a risk score of ≥ 2.5%; eligibility by B-RST™ 3.0 corresponds to a moderate or high risk score; eligibility by additional questions implies that the patient had limited family knowledge (unknown family history) or limited family structure (< 2 female members living beyond age 45 on either side)

Fig. 3

Sample images from the patient-facing PREMM₅™ application. a Sample screen inquiring about cancer history in individual relatives or small groups of relatives. b Sample screen inquiring about cancer history in the mother, which appears if the participant selects that the mother had cancer. c Sample literacy aid pop-up window depicting the family tree graphic, which appears if the patient selects the modal link titled “Who are my blood relatives?” on the screen in A. The pop-up literacy aid for sebaceous gland skin tumors that appears when the patient selects the modal link in B is depicted in Supplementary Fig. 2a

Fig. 4

Patient and provider differences in PREMM₅™ risk scores. a Scatter plot of PREMM₅™ scores derived from participant input and from genetic counselor-collected pedigrees. r_c = concordance correlation coefficient after removal two of outliers (orange, square points). Concordance correlation coefficient with the two extreme outliers included was 0.51. Magenta box = inset. b Magnified view of magenta inset in Fig. 4a, demonstrating the trend toward participant overreport on the patient-facing tool when there was disagreement between measures, as well as the excellent agreement in the two measures for some participants (points on the line of perfect concordance)