Jane Ullmer

Background/Objective: Migraine is the second most disabling condition worldwide affecting over one billion people. We aimed to investigate the frequency of recorded migraine utilising general practice and hospital admission data, comparing this with population-based prevalence estimates for Aotearoa New Zealand (NZ). Methods: From the electronic health record of de-identified data for 374,071 men and women aged 18 and over enrolled as of 2014 in NZ general practices, any entry of recorded migraine from a patient’s visit to primary care or hospital facility recorded using Read codes, Read terms, or free text entries was extracted. Via an encrypted National Health Identifier (NHI), individual patient linkage identified migraine coded in the national hospitalisation dataset, enabling exploration by gender, age group, ethnicity and NZ deprivation index quintile. Results: Compared to population prevalence (15%), 4.8% of this cohort had recorded migraine (79% in general practice only; 14% in hospital records only; 7.5% in both primary and secondary care). Women (6.7%) had higher recording than men (2.5%), with the highest age-specific recording at 45–54 years (women 7.4%, men 2.7%). By gender and ethnicity, migraine recording was highest for European women (7.9%). Within quintile 5 communities (most deprived), recording was significantly lower for females ( p < 0.0001) and males ( p = 0.03) than in quintiles 1–4. Conclusions: This study is the first to estimate migraine in NZ using linked primary and secondary care data. As well as highlighting age-, gender-, and ethnicity-specific differences, the underestimate of diagnosis compared to self-reported survey data is likely due to health care access, under-reporting, under-diagnosis, under-recording, and knowledge gaps about migraine symptoms and therapies.

Objectives: This study examined whether incorporating free-text entries into structured general practice records improves the detection of long-term conditions (LTCs) and multimorbidity (MM) in New Zealand (NZ) general practices. Methods: Data from 374 071 deidentified individuals in general practices were analysed to identify 61 LTCs. Structured data were extracted using Read codes from a national master list, and clinical raters independently identified condition-related free-text, including synonyms, negation terms and common misspellings in randomised samples. Keywords were categorised and refined through ten iterative tests. Programmatic text classification was developed and assessed against gold-standard clinician ratings, using sensitivity, specificity, positive predictive value (PPV) and F ₁ -score. Results: A quarter of general practitioner classifications contained either unrecognised Read codes or consisted of free-text only. Clinician inter-rater reliability was high (kappa ≥0.9). Compared with clinical gold standard, text classification yielded an average sensitivity of 88%, specificity of 99% and PPV of 95%, with an F ₁ -score range of 82%–95%. Incorporating free text increased LTC prevalence from 42.1% to 46.3%, reducing misclassification of MM diagnoses by identifying 12 626 additional patients with MM and 15 972 additional patients with at least one LTC. Discussion: In the course of workflow, general practitioners face barriers to accurate LTC coding or may simply annotate with text-based descriptions. Programmatic text classification has demonstrated high performance and identified many more patients receiving LTC care. Conclusions: Combining structured and unstructured data optimises MM detection in NZ general practices and has the potential to improve case management, follow-up care and allocation of healthcare resources.