Model-Informed Precision Dosing Using Machine Learning for Levothyroxine in General Practice: Development, Validation and Clinical Simulation Trial.

Levothyroxine is one of the most prescribed drugs in the western world. Dosing is challenging due to high-interindividual differences in effective dosage and the narrow therapeutic window. Model-informed precision dosing (MIPD) using machine learning could assist general practitioners (GPs), but no such models exist for primary care. Furthermore, introduction of decision-support algorithms in healthcare is limited due to the substantial gap between developers and clinicians' perspectives. We report the development, validation, and a clinical simulation trial of the first MIPD application for primary care. Stable maintenance dosage of levothyroxine was the model target. The multiclass model generates predictions for individual patients, for different dosing classes. Random forest was trained and tested on a national primary care database (n = 19,004) with a final weighted AUC across dosing options of 0.71, even in subclinical hypothyroidism. TSH, fT4, weight, and age were most predictive. To assess the safety, feasibility, and clinical impact of MIPD for levothyroxine, we performed clinical simulation studies in GPs and compared MIPD to traditional prescription. Fifty-one GPs selected starting dosages for 20 primary hypothyroidism cases without and then with MIPD 2 weeks later. Overdosage and underdosage were defined as higher and lower than 12.5 µg relative to stable maintenance dosage. MIPD decreased overdosage in number (30.5 to 23.9%, P < 0.01) and magnitude (median 50 to 37.5 µg, P < 0.01) and increased optimal starting dosages (18.3 to 30.2%, P < 0.01). GPs considered lab results more often with MIPD and most would use the model frequently. This study demonstrates the clinical relevance, safety, and effectiveness of MIPD for levothyroxine in primary care.

Who is at the centre of what? A scoping review of the conceptualisation of 'centredness' in healthcare.

OBJECTIVES: We aimed to identify the core elements of centredness in healthcare literature. Our overall research question is: How has centredness been represented within the health literature published between 1990 and 2019? METHODS: A scoping review across five databases (Medline (Ovid), PsycINFO, CINAHL, Embase (Ovid) and Scopus; August 2019) to identify all peer-reviewed literature published since 1990 that focused on the concept of centredness in any healthcare discipline or setting. Screening occurred in duplicate by a multidisciplinary, multinational team. The team met regularly to iteratively develop and refine a coding template that was used in analysis and discuss the interpretations of centredness reported in the literature. RESULTS: A total of 23 006 title and abstracts, and 499 full-text articles were screened. A total of 159 articles were included in the review. Most articles were from the USA, and nursing was the disciplinary perspective most represented. We identified nine elements of centredness: Sharing power; Sharing responsibility; Therapeutic relationship/bond/alliance; Patient as a person; Biopsychosocial; Provider as a person; Co-ordinated care; Access; Continuity of care. There was little variation in the concept of centredness no matter the preceding word (eg, patient-/person-/client-), healthcare setting or disciplinary lens. Improving health outcomes was the most common justification for pursuing centredness as a concept, and respect was the predominant driving value of the research efforts. The patient perspective was rarely included in the papers (15% of papers). CONCLUSIONS: Centredness is consistently conceptualised, regardless of the preceding word, disciplinary lens or nation of origin. Further research should focus on centring the patient perspective and prioritise research that considers more diverse cultural perspectives.

Improving the Reporting of Primary Care Research: An International Survey of Researchers.

PURPOSE: To assess opportunities to improve reporting of primary care (PC) research to better meet the needs of its varied users. METHODS: International, interprofessional online survey of PC researchers and users, 2018 to 2019. Respondents used Likert scales to rate frequency of difficulties in interpreting, synthesizing, and applying PC research reports. Free-text short answers were categorized by template analysis to record experiences, concerns, and suggestions. Areas of need were checked across existing reporting guidelines. RESULTS: Survey yielded 255 respondents across 24 nations, including 138 women (54.1%), 169 physicians (60%), 32 scientists (11%), 20 educators (7%), and 18 public health professionals (6%). Overall, 37.4% indicated difficulties using PC research reports "50% or more of the time." The most common problems were synthesizing findings (58%) and assessing generalizability (42%). Difficulty was reported by 49% for qualitative, 46% for mixed methods, and 38% for observational research. Most users wanted richer reporting of theoretical foundation (53.7%); teams, roles, and organization of care (53.4%); and patient involvement in the research process (52.7%). Few reported difficulties with ethics or disclosure of funding or conflicts. Free-text answers described special challenges in reporting PC research: context of clinical care and setting; practical details of interventions; patient-clinician and team relationships; and generalizability, applicability and impact in the great variety of PC settings. Cross-check showed that few current reporting guidelines focus on these needs. CONCLUSIONS: Opportunities exist to improve the reporting of PC research to make it more useful for its many users, suggesting a role for a PC research reporting guideline.

Exploring the psychometric properties of the Working Alliance Inventory in general practice: a cross-sectional study.

BACKGROUND: The therapeutic alliance is a framework from psychology that describes three components: goals, tasks, and bond. The Working Alliance Inventory adapted for general practice (WAI-GP) measures the strength of the therapeutic alliance between the patient and the clinician, and it could be useful in both research and clinical settings. AIM: To determine if the patient score on WAI-GP can delineate the three components (goals, tasks, and bond), and to test concurrent validity with the Consultation and Relational Empathy (CARE) measure and the Patient Perception of Patient-Centredness (PPPC) measure. DESIGN & SETTING: A cross-sectional study took place in 12 general practice waiting rooms in Australia. METHOD: The research instruments included the 12-item WAI-GP (the patient version), the CARE and PPPC measures, plus a survey of demographics and reason for consultation. To perform a principal components factor analysis of the WAI-GP, this dataset was combined with an existing dataset. The Spearman rank correlation was used to determine concurrent validity between the WAI-GP and the CARE and PPPC measures. RESULTS: Participants (97-99%) reported a strong positive alliance after the consultation (average WAI-GP mean 4.27 ± 0.67 out of 5, n = 146). Factor analysis could not separate the three components (one factor, eigenvalue >1; Cronbach's α = 0.957; n = 281). Concurrent validity was supported by moderate correlations with the other measures (PPPC ρ = -0.51, P<0.005, CARE ρ = 0.56, P<0.005). CONCLUSION: Three components could not be identified, but the WAI-GP has a high internal consistency and concurrent validity with moderate correlations with the CARE and PPPC. A more diverse sample may better distinguish the three components leading to more specific feedback to clinicians on their consultation practices.