Fiji is a Pacific Island nation with the predominant ethnic groups indigenous Fijians (iTaukei) (62 %) and Fijians of Indian descent (31 %). This study reports on the effect of a Parental Assistance Payment Program (PAPP) tied to on-time birth registration, available in Fiji from August 2018 to July 2020.
Unit record birth registration data (n = 117,829) for children born during 2016–22 were used to calculate mean birth-to-registration intervals and the likelihood of on-time birth registration (within 365 days) before the PAPP (January 2016–July 2018) compared to during the PAPP (August 2018–July 2020), by population disaggregations (sex, ethnicity, age, marital status).
During the PAPP, mean birth-to-registration intervals declined sharply by 81 %, from 665 days (95 %CI: 658–671) to 124 days (121–127). The largest declines were among i-Taukei children (803 to 139 days, 83 %) compared to non-iTaukei (283 to 76 days, 73 %); mothers aged 10–19 years (880 to 134 days, 85 %) compared to ≥20 years (653 to 123 days, 81 %); and single mothers (983 to 145 days, 85 %) compared to married mothers (570 to 115 days, 80 %). On-time birth registration increased from 57 % to 93 %, and the adjusted hazard ratio showed children born during the PAPP were 2.3 times more likely (95 %CI: 2.2–2.4) to have their birth registered on-time compared to children born before the PAPP. When the PAPP was discontinued in August 2020, the birth-to-registration interval increased sharply in all population groups.
During the two-year period the PAPP was available, it was highly effective at improving the timeliness of birth registration, particularly among iTaukei children, young mothers, and single mothers. After the PAPP was discontinued, the timeliness of birth registration deteriorated sharply. Longer post-PAPP follow-up time (≠5 years) is required to determine whether the timeliness of birth registration has deteriorated to levels similar to those during the pre-PAPP period.
The successful implementation and interpretation of machine learning (ML) models in epidemiological studies can be challenging without an extensive programming background. We provide a didactic example of machine learning for risk prediction in this study by determining whether early life factors could be useful for predicting adolescent psychopathology.
In total, 9643 adolescents ages 9–10 from the Adolescent Brain and Cognitive Development (ABCD) Study were included in ML analysis to predict high Child Behavior Checklist (CBCL) scores (i.e., t-scores ≥ 60). ML models were constructed using a series of predictor combinations (prenatal, family history, sociodemographic) across 5 different algorithms. We assessed ML performance through sensitivity, specificity, F1-score, and area under the curve (AUC) metrics.
A total of 1267 adolescents (13.1 %) were found to have high CBCL scores. The best performing algorithms were elastic net and gradient boosted trees. The best performing elastic net models included prenatal and family history factors (Sensitivity 0.654, Specificity 0.713; AUC 0.742, F1-score 0.401) and prenatal, family, history, and sociodemographic factors (Sensitivity 0.668, Specificity 0.704; AUC 0.745, F1-score 0.402). Across all 5 ML algorithms, family history factors (e.g., either parent had nervous breakdowns, trouble holding jobs/fights/police encounters, and counseling for mental issues) and sociodemographic covariates (e.g., maternal age, child's sex, caregiver income and caregiver education) tended to be better predictors of adolescent psychopathology. The most important prenatal predictors were unplanned pregnancy, birth complications, and pregnancy complications.
Our results suggest that inclusion of prenatal, family history, and sociodemographic factors in ML models can generate moderately accurate predictions of adolescent psychopathology. Issues associated with model overfitting, hyperparameter tuning, and system seed setting should be considered throughout model training, testing, and validation. Future early risk predictions models may improve with the inclusion of additional relevant covariates.
The use of unverified models for risk estimates and policy recommendations can be highly misleading, as their predictions may not reflect real-world health impacts. For example, a recent article states that NO2 from gas stoves “likely causes ∼50,000 cases of current pediatric asthma from long-term NO2 exposure alone” annually in the United States. This explicitly causal claim, which is contrary to several methodology and review articles published in this journal, among others, reflects both (a) An unverified modeling assumption that pediatric asthma burden is approximately proportional to NO2; and (b) An unverified causal assumption that the assumed proportionality between exposure and response is causal. The article is devoid of any causal analysis showing that these assumptions are likely to be true. It does not show that reducing NO2 exposure from gas stoves would reduce pediatric asthma risk. Its key references report no significant associations – let alone causation – between NO2 and pediatric asthma. Thus, the underlying data suggests that the number of pediatric asthma cases caused by gas stoves in the United States is indistinguishable from zero. This highlights the need to rigorously validate modeling assumptions and causal claims in public health risk assessments to ensure scientifically sound foundations for policy decisions.
Autoimmune diseases account for a substantial burden of disease in high-income countries, including Europe and North America. However, their epidemiology remains under-researched in other regions. We examined the incidence and prevalence of eight autoimmune diseases in the adult Chinese population through a systematic review of epidemiological studies.
We searched OvidSP MEDLINE and Google Scholar from 1995 to 2023 (inclusive) for articles on the incidence or prevalence of autoimmune thyroiditis (AT), Graves' disease (GD), type 1 diabetes mellitus (T1D), multiple sclerosis (MS), Crohn's disease (CD), ulcerative colitis (UC), rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE). We included studies from mainland China, Taiwan, Hong Kong or Macau. The study is registered with PROSPERO (CRD42021225842).
We retrieved 2278 records, of which 62 studies (161 estimates) were included in the systematic review, and 42 studies (101 estimates) were included in the meta-analysis. Pooled fixed-effects estimates for incidence of CD, UC, MS, T1D and SLE were 0.22 (95% CI 0.21–0.23), 1.13 (1.10–1.17), 0.28 (0.26–0.30), 2.20 (1.70–2.84) and 4.87 (4.21–5.64) per 100,000 persons, respectively. For RA, one study estimate was included, with an incidence of 15.8 per 100,000 persons. Fixed-effects estimates for the prevalence of CD, UC, MS, SLE, RA, GD and AT were 3.73 (95% CI 3.68–3.78), 16.11 (15.93–16.29), 4.08 (3.95–4.21), 93.44 (92.27–94.63), 104 (103–106), 450 (422–481) and 2322 (2057-2620), respectively, per 100,000 persons. Across all conditions, women were almost twice as likely as men to be diagnosed with an autoimmune disease.
There is marked variation in the frequency of autoimmune diseases among Chinese adults. We estimate that 2.7–3.0% (>31 million people) of the adult Chinese population have one or more autoimmune diseases, comparable to Western populations, with the majority of the burden from autoimmune thyroid diseases and rheumatoid arthritis.
The healthcare system in Ireland was profoundly affected by COVID-19. This study aimed to explore the impact of the pandemic on cancer surgery in Ireland, from 2019 to 2022 using three national health data sources.
A repeated cross-sectional study design was used and included: (i) cancer resections from the National Histopathology Quality Improvement (NHQI) Programmes; (ii) cancer surgery from the National Cancer Registry Ireland (NCRI), and (iii) cancer surgery from Hospital Inpatient Enquiry (HIPE) System. Cancer surgery was presented by invasive/in situ and invasive only cancers (NCRI & HIPE), and by four main cancer types (breast, lung, colorectal & melanoma for NCRI & HIPE data only).
The annual number of cancer resections (NHQI) declined by 4.4% in 2020 but increased by 4% in 2021 compared with 2019. NCRI data indicated invasive/in-situ cancer surgery for the four main cancer types declined by 14% in 2020 and 5.1% in 2021, and by 12.3% and 7.3% for invasive cancer only, compared to 2019. Within HIPE for the same tumour types, invasive/in situ cancer surgery declined by 21.9% in 2020 and 9.9% in 2021 and by 20.8% and 9.6% for invasive cancer only. NHQI and HIPE data indicated an increase in the number of cancer surgeries performed in 2022.
Cancer surgery declined in the initial pandemic waves suggests mitigation measures for cancer surgery, including utilising private hospitals for public patients, reduced the adverse impact on cancer surgery.
Sustainable Development Goal 3.4.1 (SDG3.4.1) targets a one-third reduction in non-communicable disease (NCD) mortality in ages 30–69-years by 2030 (relative to 2015). Directing interventions to achieve this aim requires reliable estimates of underlying cause of death (UCoD). This may be problematic when both cardiovascular diseases (CVD) and diabetes are present due to a lack of consistency in certification of such deaths. We estimate empirically 2013–17 NCD mortality in Fiji, by sex and ethnicity, from CVD, diabetes, cancer, and chronic lower respiratory diseases (CRD), and aggregated as NCD4.
UCoD was determined from Medical Certificates of Cause-of-Death (MCCD) from the Fiji Ministry of Health after pre-processing of mortality data where diabetes and/or hypertension were present in order to generate internationally comparable UCoD. If no potentially fatal complications from diabetes or hypertension accompanied these causes in Part I (direct cause) of the MCCD, these conditions were re-assigned to Part II (contributory cause). The probability of a 30-year-old dying before reaching age 70-years (PoD30–70), by cause, was calculated.
The PoD30–70 from NCD4 over 2013–17 differed by sex and ethnicity: in women, it was 36% (95%CI 35–37%) in i-Taukei and 27% (26–28%) in Fijians of Indian descent (FID); in men, it was 41% (40–42%) in both i-Taukei and FID.
PoD30–70 from CVD, diabetes, cancer and CRD in women was: 18%, 10%, 13% and 1·0% in i-Taukei; 13%, 10%, 5·6% and 1·1% in FID; in men was: 28%, 8.4%, 7·6% and 2·2% in i-Taukei; 31%, 8.3%, 3.5% and 3·1% in FID.
To achieve SDG3.4.1 goals in Fiji by 2030, effective population wide and ethnic-specific interventions targeting multiple NCDs are required to reduce PoD30–70 from NCD4: from 36% to 24% in i-Taukei, and 27% to 18% in FID women; and from 41% to 27% in i-Taukei and FID men.
Not applicable.
Bias away from the null in odds ratios (OR), aggravated by low power, is a well-known phenomenon in statistics (sparse data bias). Such bias increases in presence of selection of “significant” results on the basis of null hypothesis testing (effect size magnification, ESM).
We seek to illustrate these issues and adjust for suspected sparse data bias in the context of a reported more than doubling of the odds of new onset type 2 diabetes in presence of occupational trichlorfon insecticide exposure reported in the Agricultural Health Study.
We performed ESM analysis on the crude ORs extracted from the contingency table in the published report, which is done by simulating selected OR given a posited true OR. Next, we applied easily accessible methods that adjust for sparse data bias to the extracted contingency tables, including data augmentation, bootstrap, Firth's regression, and Bayesian methods with weakly informative priors.
During the ESM analysis, we observed that there was a reasonable chance that a “statistically significant” OR of around 2.5–2.6 would be observed for true OR of 1.2. Adjustment for sparse data bias revealed that Bayesian methods outperformed alternative approaches in terms of yielding more precise inference, while not making unjustified distributional assumptions about estimates of OR. The OR in the original paper of about 2.5–2.6 was reduced on average to OR of 1.9 to 2.2, with 95% (Bayesian) credible intervals that included the null.
It is reasonable to adjust ORs for sparse data bias when the reported association has societal importance, because policy must be informed by the least biased estimates of the effect. We think that such adjustment would lead to a more appropriate evaluation of the extent of evidence on the contribution of occupational exposure to trichlorfon pesticide to risk of new onset diabetes.
Uncovering the root causes of complex diseases requires complex approaches, yet many studies continue to isolate the effects of genetic and social determinants of disease. Epidemiologic efforts that under-utilize genetic epidemiology methods and findings may lead to incomplete understanding of disease. Meanwhile, genetic epidemiology studies are often conducted without consideration of social and environmental context, limiting the public health impact of genomic discoveries. This divide endures despite shared goals and increases in interdisciplinary data due to a lack of shared theoretical frameworks and differing language. Here, we demonstrate that bridging epidemiological divides does not require entirely new ways of thinking. Existing social epidemiology frameworks including Ecosocial theory and Fundamental Cause Theory, can both be extended to incorporate principles from genetic epidemiology. We show that genetic epidemiology can strengthen, rather than detract from, efforts to understand the impact of social determinants of health. In addition to presenting theoretical synergies, we offer practical examples of how genetics can improve the public health impact of epidemiology studies across the field. Ultimately, we aim to provide a guiding framework for trainees and established epidemiologists to think about diseases and complex systems and foster more fruitful collaboration between genetic and traditional epidemiological disciplines.