Statistics Feed

## The Finding The night sky is getting brighter at a rate of approximately **9.6% per year** as experienced by human observers — far faster than satellite-based measurements had suggested. At this pace, a child born under skies where 250 stars were visible would see fewer than 100 by age 18. This rate implies a **doubling of sky brightness roughly every 8 years**. ## Who Conducted the Research The study was led by **Christopher Kyba** of the German Research Centre for Geosciences (GFZ) and the Ruhr-Universität Bochum, along with collaborators from the NOIRLab (U.S. National Science Foundation) and other institutions. It was published in the journal **Science** in January 2023. ## Methodology The researchers analysed data from **Globe at Night**, a long-running citizen science programme. Participants go outside at night and compare their view of specific constellations against a set of reference charts showing progressively fainter stars. Each chart corresponds to a **naked-eye limiting magnitude** (NELM) — the faintest star visible. This provides a standardised, human-calibrated measure of sky brightness at each observer's location. The dataset comprised **51,351 individual observations** collected between **2011 and 2022**. Observations were filtered to exclude those taken under cloudy skies, during moonlit periods, or from locations with incomplete metadata. The team then modelled the change in NELM over time, controlling for observer location and other confounders. ## Geographic Scope and Demographics Participants were located across the globe, but the dataset was **heavily weighted toward North America and Europe**, which together accounted for the large majority of observations. Observers ranged from students to amateur astronomers — essentially a **self-selected convenience sample** of people interested in night-sky quality. ## Why It Diverges from Satellite Data A critical insight is that satellite instruments (notably the VIIRS Day-Night Band on the Suomi NPP satellite) had estimated light-pollution growth at roughly **2% per year globally**. The roughly fivefold discrepancy arises because satellites cannot detect certain wavelengths — particularly the **blue-rich emissions from modern LED lighting**, which scatter efficiently in the atmosphere and strongly affect human night vision. Satellites also struggle with light emitted at very low angles (e.g., illuminated building facades, signage). The study thus reveals that satellites have been substantially **underestimating the pace of skyglow growth**. ## Key Limitations - **Self-selection bias**: Globe at Night participants are not a random population sample; they are people motivated to observe the sky, potentially from darker locations, which could skew trends. - **Observer consistency**: Individual ability to detect faint stars varies with age, visual acuity, and dark-adaptation time. The study relies on aggregate trends to smooth out individual variability, but systematic biases remain possible. - **Geographic bias**: The dominance of North American and European observations means the 9.6% figure may not be globally representative. Light-pollution trajectories in sub-Saharan Africa, South Asia, and South America are undersampled. - **Coarse measurement scale**: NELM is reported in discrete steps (whole or half magnitudes), limiting precision at the individual-observation level. - **Confounders**: Changes in atmospheric aerosol levels, wildfire smoke, or humidity patterns over the study period could influence perceived sky brightness independently of artificial light. ## Interpretation This finding does **not** mean every location on Earth is losing stars at the same rate — local trends depend on urban development, lighting policies, and geography. It **does** demonstrate that the human experience of light pollution is worsening far more rapidly than satellite monitoring would indicate, with implications for **circadian health, nocturnal ecosystems, astronomical observation, and cultural heritage**. The study has been widely cited in subsequent work on LED lighting policy and has reinforced calls for better ground-based monitoring networks alongside satellite surveillance.

## The Finding In 2019, bacterial antimicrobial resistance (AMR) was directly responsible for an estimated **1.27 million deaths** worldwide and associated with **4.95 million deaths** in total. This made drug-resistant bacterial infections a leading global cause of death — killing more people than HIV/AIDS (864,000 deaths that year) or malaria (643,000 deaths). ## Who Conducted the Research The study was carried out by the **Antimicrobial Resistance Collaborators**, a large consortium coordinated by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington, led by Prof. Christopher J.L. Murray. It was published in ***The Lancet*** in January 2022. ## Methodology This was a **comprehensive systematic analysis** that synthesised data from multiple source types: - Hospital and laboratory surveillance systems - Pharmaceutical sales data - Published literature (systematic review) - Microbiology datasets from individual hospitals and networks - Cause-of-death data from vital registration systems and verbal autopsies The researchers built a series of **counterfactual statistical models** to estimate two quantities: (1) deaths *directly attributable* to AMR (comparing outcomes when a resistant infection is present versus a drug-susceptible infection), and (2) deaths *associated* with AMR (comparing resistant infections versus no infection at all). They modelled **23 bacterial pathogens**, **88 pathogen–drug combinations**, and **11 infectious syndromes** across **204 countries and territories**. Where primary data were unavailable — particularly in low-income countries — the team used Bayesian predictive models and spatial–temporal Gaussian process regression to generate estimates. ## Sample and Scope The analysis drew on **471 million individual records** and data points, including microbiology data from 7,585 study–location–year combinations. The geographic scope was global, covering all WHO regions, with sub-Saharan Africa and South Asia bearing the highest attributable mortality rates. ## Time Period All estimates refer to the calendar year **2019**, chosen to avoid confounding from the COVID-19 pandemic. ## Key Pathogens Six pathogens were responsible for the majority of attributable deaths: 1. *Escherichia coli* — 157,082 deaths 2. *Staphylococcus aureus* (particularly MRSA) — 121,252 deaths 3. *Klebsiella pneumoniae* — 193,139 deaths 4. *Streptococcus pneumoniae* — 141,538 deaths 5. *Acinetobacter baumannii* — 94,085 deaths 6. *Pseudomonas aeruginosa* — 78,579 deaths ## Limitations - **Data sparsity in low-income settings**: Many countries in sub-Saharan Africa and South Asia lack robust microbiology surveillance, so estimates for these regions rely heavily on statistical modelling and extrapolation, introducing considerable uncertainty. - **Counterfactual assumptions**: The attributable mortality estimate depends on the assumption that a drug-susceptible infection would have replaced the resistant one. This is debatable — some patients might not have been infected at all. - **Hospital bias**: Microbiology data overwhelmingly come from hospital settings, potentially underrepresenting community-acquired resistant infections or overrepresenting severe cases. - **No viral or fungal AMR**: The study focused exclusively on bacterial resistance, excluding antifungal and antiviral resistance. - **Cross-sectional design**: A single year of data cannot establish trends over time. ## Interpretation The finding establishes AMR as one of the most significant global health threats, rivalling or exceeding major infectious diseases in mortality burden. However, it does **not** mean 1.27 million people died solely because antibiotics failed — many had complex comorbidities and the infections themselves were often the proximate cause. The figure represents the *additional* deaths caused by the resistance rather than the infection alone. The study was a landmark in quantifying a problem that had previously been described largely through projections, and it has since informed WHO and G7 policy priorities on AMR stewardship.

## The Finding In 2019, bacterial antimicrobial resistance (AMR) was associated with an estimated **4.95 million deaths** worldwide, of which **1.27 million deaths** were *directly attributable* to resistant infections — meaning those people would not have died had the infections been drug-susceptible. This makes AMR one of the leading causes of death globally, exceeding HIV/AIDS (~860,000 deaths) and malaria (~640,000 deaths) that year. ## Who Conducted the Research The study was carried out by the **Antimicrobial Resistance Collaborators**, a large consortium operating within the Institute for Health Metrics and Evaluation (IHME) Global Burden of Disease framework. It was published in **The Lancet** in January 2022. ## Methodology This was a **comprehensive systematic analysis** combining multiple data sources and statistical models: - Researchers assembled **471 million individual records or isolates** from systematic literature reviews, hospital systems, surveillance networks, and other data sources across **204 countries and territories**. - They examined **23 bacterial pathogens** and **88 pathogen–drug combinations**. - The analysis employed a **counterfactual modeling framework** with two key scenarios: (1) deaths attributable to AMR (comparing outcomes of drug-resistant infections vs. drug-susceptible infections), and (2) deaths associated with AMR (comparing outcomes of drug-resistant infections vs. no infection at all). - Bayesian statistical models were used to estimate pathogen distributions, resistance proportions, and mortality where direct data were sparse — particularly in low- and middle-income countries. ## Geographic Scope and Key Patterns The burden was **global but unevenly distributed**. Sub-Saharan Africa and South Asia bore the highest rates of directly attributable AMR deaths, with western sub-Saharan Africa experiencing roughly **27.3 attributable deaths per 100,000 population**. In contrast, Australasia had about **6.5 per 100,000**. Six leading pathogens — *Escherichia coli*, *Staphylococcus aureus*, *Klebsiella pneumoniae*, *Streptococcus pneumoniae*, *Acinetobacter baumannii*, and *Pseudomonas aeruginosa* — accounted for **929,000** of the 1.27 million directly attributable deaths. ## Time Period All estimates pertain to the calendar year **2019**, deliberately chosen to avoid confounding by the COVID-19 pandemic. ## Key Limitations - **Data sparsity in low-income settings**: Many countries in sub-Saharan Africa and South Asia lacked high-quality microbiology laboratory data, requiring heavy reliance on statistical extrapolation. - **Hospital-based sampling bias**: Much of the data came from hospital surveillance systems, which may over-represent severe cases and under-represent community-acquired infections. - **Counterfactual assumptions**: The two counterfactual scenarios involve inherently untestable assumptions — particularly the "no infection" scenario, which likely overstates the role of resistance alone. - **Pathogen scope**: Only bacterial AMR was modeled; resistance in fungi, viruses, and parasites (e.g., drug-resistant malaria or tuberculosis) was excluded. - **Temporal snapshot**: This is a single-year estimate without trend data, making it impossible to determine whether the burden is rising or stable from this study alone. ## Interpretation The 1.27 million figure represents the most conservative estimate — deaths *directly caused* by resistance. The 4.95 million figure captures broader associated mortality, including deaths where resistance played a contributing role. Neither figure should be interpreted as meaning AMR "killed" nearly 5 million people in isolation; rather, resistance worsened outcomes of infections that themselves carried mortality risk. The study powerfully demonstrates that AMR is not a future hypothetical but a **present-day crisis** requiring urgent investment in diagnostics, stewardship, new antibiotics, infection prevention, and especially laboratory infrastructure in underserved regions.

## The Finding The total breeding bird population in the United States and Canada has declined by an estimated **2.9 billion individuals** since 1970, representing a net loss of **29%** of the total avian abundance. The steepest losses occurred among common species in grassland habitats, which experienced a **53% decline** (more than 700 million birds), but the pattern extended across nearly every major biome including forests, which lost over 1 billion birds. ## Who Conducted the Research The study was led by **Kenneth V. Rosenberg** of the Cornell Lab of Ornithology and the American Bird Conservancy, alongside collaborators from institutions including Environment and Climate Change Canada, the U.S. Geological Survey, the Smithsonian Conservation Biology Institute, and others. It was published in the journal **Science** in September 2019. ## Methodology The researchers used a **multi-dataset, convergent-evidence approach** drawing on three independent monitoring sources: 1. **The North American Breeding Bird Survey (BBS)** — a standardized, long-running roadside survey coordinated by the USGS and Canadian Wildlife Service since 1966. Skilled volunteer observers conduct 3-minute point counts at 50 stops along ~4,100 randomly placed routes each June. Hierarchical models were applied to estimate annual population trajectories for **529 species**. 2. **The Audubon Christmas Bird Count (CBC)** — the longest-running citizen science bird survey, providing winter abundance data since 1900, used here as a complementary dataset to validate BBS trends. 3. **NEXRAD weather radar data** — a novel approach using the national network of 143 Doppler weather stations to measure the **biomass of nocturnally migrating birds** in spring passage over the continental U.S. from 2007 to 2017. This provided a fully independent, technology-based cross-check unaffected by observer effort biases. Total population sizes were estimated by combining BBS-derived relative abundance indices with **Partners in Flight population size estimates**, which use density correction factors calibrated against intensive studies. ## Scope and Population - **Geographic scope:** United States and Canada - **Species covered:** 529 native bird species with sufficient BBS monitoring data, representing over 75% of all breeding landbird species on the continent - **Time period:** 1970–2017 for survey data; 2007–2017 for radar data - **Excluded:** Rare, nocturnal, or colonial species poorly sampled by roadside surveys, as well as non-native species ## Key Limitations - **BBS roadside bias:** Routes are placed along roads, which may under-represent deep-forest or remote-habitat species and over-represent edge-dwelling species. - **Population estimation uncertainty:** Converting relative abundance indices into absolute population counts requires density multipliers that carry substantial uncertainty. - **Radar data temporal scope:** The independent radar analysis covers only 2007–2017, a relatively short window, though it corroborated the declining trajectory. - **Causation not established:** The study documents the decline but does not directly attribute it to specific drivers, though habitat loss, agricultural intensification, pesticide use, cat predation, window collisions, and climate change are all discussed as plausible contributors. - **Self-selection of volunteer observers:** Both BBS and CBC rely on skilled volunteers, introducing possible effort and detection biases over time, though statistical models attempt to account for this. ## Interpretation This finding means that roughly **one in every four birds** present in North America in 1970 is now gone. Critically, the losses are not confined to already-rare or endangered species — they are overwhelmingly driven by declines in **common, widespread species** such as sparrows, warblers, blackbirds, and finches. This pattern suggests broad-scale environmental degradation rather than species-specific threats. The convergence of survey data and radar data strengthens the finding considerably. However, the study does not prove any single cause and should not be interpreted as forecasting inevitable extinction. Some groups — notably raptors and waterfowl — have **increased** over the same period, likely due to targeted conservation efforts like the banning of DDT and wetland protection, demonstrating that policy interventions can reverse declines. ## Related Findings Similar continental-scale declines have been documented in **European farmland birds** (56% decline, 1980–2018, per the Pan-European Common Bird Monitoring Scheme) and in **insect biomass** in Europe, suggesting a broader pattern across taxa in human-dominated landscapes.

## The Finding In a 2022 study published in the journal *Environment International*, researchers detected quantifiable levels of microplastic particles in the blood of **17 out of 22 healthy adult volunteers (77%)**. The mean total concentration of plastic particles across all positive samples was **1.6 µg/mL of whole blood**. The most frequently detected polymers were PET (polyethylene terephthalate, found in 50% of donors), polystyrene (36%), and polyethylene (23%). ## Who Conducted the Research The study was led by **Heather A. Leslie** and colleagues at Vrije Universiteit Amsterdam, in collaboration with the Amsterdam University Medical Center and several other Dutch institutions. It was published in *Environment International* (Volume 163, May 2022), a peer-reviewed Elsevier journal focused on environmental health sciences. ## Methodology ### Study Design & Instruments This was a **cross-sectional analytical study** employing a novel double-shot pyrolysis–gas chromatography/mass spectrometry (Py-GC/MS) method. This technique thermally decomposes plastic polymers into signature chemical fragments, which are then identified and quantified by mass spectrometry. The method was specifically developed and validated to detect five major polymer types (PET, polyethylene, polystyrene, polypropylene, and polymethyl methacrylate) at low microgram-per-milliliter concentrations in a complex biological matrix like blood. ### Sample Size & Sampling The sample consisted of **22 healthy adult volunteers** recruited from the general population in the **Netherlands**. This was a **convenience sample** — donors were anonymous, and limited demographic information was collected (all were adults aged ≥18). Blood was drawn using steel needles and glass syringes to minimise plastic contamination, and extensive procedural blanks were run alongside every batch. ### Time Period Blood samples were collected in **2021**, with the study published in March 2022. ## Key Limitations - **Very small sample size (n=22):** The study was primarily a proof-of-concept to demonstrate that microplastics *can* be detected and quantified in human blood. It was never designed to estimate population-level prevalence. - **Convenience sampling:** Volunteers were not randomly selected, so the 77% detection rate cannot be generalised to the broader Dutch or global population. - **Contamination risk:** Despite rigorous blank controls and use of non-plastic collection equipment, background contamination is an inherent challenge in microplastics research. The authors subtracted blank values, but trace contamination can never be entirely ruled out. - **Snapshot measurement:** A single blood draw captures only circulating particles at one moment — it tells us nothing about long-term exposure, accumulation in organs, or health consequences. - **No health outcome data:** The study did not link microplastic concentrations to any clinical endpoints. Detection in blood does not, by itself, demonstrate harm. - **Funding:** The work was funded partly by Common Seas, an advocacy organisation working to reduce plastic pollution, which could represent a potential conflict of interest, though the analytical methods were peer-reviewed. ## Interpretation This study provided the **first direct evidence** that plastic particles can enter the human bloodstream and be analytically measured there. It demonstrates systemic bioavailability — meaning microplastics are not just passing through the gut but are reaching internal compartments. However, it does **not** prove that these concentrations cause disease. Toxicological significance at these levels remains an open research question. ## Replications & Related Work Subsequent studies have reinforced the plausibility of these findings. A 2024 study in the *New England Journal of Medicine* (Marfella et al.) detected microplastics in **carotid artery plaque** tissue and associated their presence with elevated cardiovascular risk, though causality was not established. Other groups have detected microplastics in lung tissue, placenta, and breast milk, collectively building a picture of widespread human internal exposure.

## The Finding A 2022 study published in the journal *Environment International* reported that quantifiable levels of plastic particles were found in **17 out of 22 (77%)** blood samples collected from healthy adult volunteers. The mean total concentration of plastic particles in blood was **1.6 µg/mL**, with some individual samples reaching concentrations several times higher. ## Who Conducted the Research The study was led by Heather A. Leslie and colleagues at the Department of Environment and Health, Vrije Universiteit Amsterdam, in the Netherlands. It was published in May 2022 (Volume 163, Article 107199) and has since become one of the most-cited papers in the emerging field of human microplastic exposure. ## Methodology The researchers developed and validated an **analytical method using double-shot pyrolysis–gas chromatography/mass spectrometry (Py-GC/MS)** to identify and quantify plastic particles ≥700 nm in whole human blood. This technique thermally decomposes polymers and identifies them by their characteristic molecular fragments. To control for background contamination — a critical challenge in microplastics research — the team used **steel syringe needles**, pre-washed glassware, and procedural blanks processed identically to real samples. Results were blank-corrected to subtract any contamination introduced during laboratory handling. ## Sample and Population The sample comprised **22 healthy adult volunteers** recruited from the general population in the Netherlands. This was a **convenience sample**, not a random or stratified selection. Donors were anonymized, and limited demographic data were collected. Blood was drawn via venipuncture into glass heparinized tubes. ## Key Results by Polymer Type - **PET (polyethylene terephthalate):** detected in 50% of samples - **Polystyrene:** detected in 36% of samples - **Polyethylene:** detected in 23% of samples - **Poly(methyl methacrylate):** detected in 5% of samples These are among the most commonly produced plastics worldwide, found in packaging, textiles, food containers, and disposable products. ## Limitations and Sources of Bias 1. **Very small sample size (n = 22):** This limits generalizability and statistical power. Population-level prevalence estimates cannot be reliably drawn from this study alone. 2. **Convenience sampling:** Participants were not randomly selected, introducing potential selection bias. 3. **Single time-point measurement:** Blood was drawn once per person, so the study cannot speak to temporal variation, accumulation, or clearance rates. 4. **Detection threshold:** Only particles ≥700 nm were quantifiable; smaller nanoplastics — potentially more biologically active — were below the limit of detection. 5. **Contamination risk:** Despite rigorous blank correction, microplastics research is inherently vulnerable to environmental contamination during sample collection and processing. 6. **Health effects unknown:** The study measured *presence* and *concentration* only. It made **no claims** about health outcomes or toxicological effects at the concentrations observed. ## Interpretation This study provided the **first direct evidence** that plastic particles are present in human blood, suggesting systemic distribution after exposure via ingestion, inhalation, or dermal contact. However, it does **not** demonstrate that these concentrations cause harm. The toxicological significance of blood-borne microplastics at low µg/mL levels remains an open research question. ## Replications and Related Work Subsequent studies have corroborated the widespread presence of microplastics in human tissues. A 2024 study in the *New England Journal of Medicine* (Marfella et al.) detected microplastics in carotid artery plaque and found an association with cardiovascular events, though causality was not established. Research by Jenner et al. (2022) also detected microplastics deep in human lung tissue. The field is rapidly evolving, with ongoing efforts to standardize detection methods and establish dose-response relationships.

# Test