Estimating annual deaths from stroke in adults under 70 years of age in Freetown Sierra Leone: A comparative analysis of a hospital-based stroke register and a population-based verbal autopsy study
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Background: In Sub-Saharan Africa (SSA), most stroke epidemiological data comes from hospital-based registers, which are prone to selection bias, and data may be unrepresentative of stroke burden at the population level. The degree of incompleteness and bias in hospital-based registers has been assessed in high-income countries but not in an SSA country. Aims: The study describes and compares estimates of annual deaths from stroke under 70 years of age, from a hospital-based stroke register and a population-based verbal autopsy (VA) study. We describe the sociodemographic and clinical differences between patients captured and those missed by a hospital-based register and estimate the completeness of a hospital-based register in Sierra Leone. Methods: We compared people under 70 years of age who died from stroke in the Stroke in Sierra Leone (SISLE) prospective longitudinal hospital-based register to the Healthy Sierra Leone (HEAL-SL) population-based VA study which sampled 2.5% of households in the Western Area. We included participants from SISLE and HEAL-SL who died within the same dates (1st May 2019 until 30th September 2021) and geographical area. We conducted data linkage using probabilistic matching and manual clerical review by two authors. To assess selection bias, we used univariable analysis to identify variables associated with capture by the hospital register. To estimate annual deaths from stroke, two-source capture-recapture analysis was conducted using the Lincoln-Petersen-Chapman estimator. Estimates of completeness were adjusted for undermatching and for the positive predictive value of VA for stroke diagnosis. Deaths rates from stroke were calculated as deaths per 100,000 individuals, with population estimates sourced from the 2021 Mid-term Population and Housing Census. Results: A total of 345 participants were identified in the SISLE dataset, 46 in the VA dataset, and 4 in both datasets. Excluding individuals captured in both datasets, individuals identified by VA had a mean age of 58 years compared to 55 years in SISLE ( p = 0.07), 59.5% were male compared to 50.7% in SISLE ( p = 0.28), and 52.3% had no formal education compared to 39.0% ( p = 0.09) in SISLE. Individuals identified by VA were more likely to be employed 36.7% vs 59.5% ( p = 0.002), were less likely to have sought formal healthcare 48.5% vs 100% ( p < 0.001), more likely to have died suddenly 14.3% vs 4.1% ( p < 0.001), and less likely to have died in hospital 19.0% vs 67.5%. Estimates of annual deaths from stroke using capture-recapture methods ranged from 41 to 106/100,000. The completeness of SISLE register for fatal stroke ranged from 10.6% (95% CI: 9.6%–11.7%) to 27.2% (95% CI: 24.8%–30.0%). Discussion: In this setting, a hospital-based stroke register underestimated deaths from stroke in adults younger than 70 years to a much greater degree than estimates from high-income country settings. For people who died from SISLE, employed people, people who did not seek formal healthcare, and people who died within 24 hours were less likely to be included in the hospital-based stroke register. Investment in routine death registration systems and population-based stroke surveillance is essential to provide accurate estimates of population-level stroke burden in our setting.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it