Introduction

Background

For more than 40% of the global human population who live in drylands, blowing dust from deserts or agricultural lands is an important component of the air pollution burden they bear. The health effects related to windblown dust exposure and their potential causal mechanisms have been documented in numerous studies and reviews (e.g., Aghababaeian et al. 2021; Fussell and Kelly 2021; Goudie 2014; Lwin et al. 2023; Middleton 2020; Sadeghimoghaddam et al. 2021; Tong et al. 2023; Zhang et al. 2016). Mineral/soil dust has been associated most commonly with respiratory and cardiovascular diseases but also with infectious diseases, eye problems, skin disease, cerebrovascular diseases, adverse birth outcomes, allergic diseases, accidents and injuries, diseases associated with maternity and reproduction, mental health, diabetes, and other health conditions (Lwin et al. 2023). However, the majority of these studies (e.g., Hashizume et al. 2020; Middleton 2020) have taken place in the “global dust belt” (Prospero et al. 2002), which extends from West Africa across the Middle East to northeast Asia.

Fewer population-based studies of the health effects of windblown dust have been performed in North America (Tong et al. 2023). For example, Gomez et al. (1992) found an increased prevalence of cough, wheeze, and eye and nasal irritation in persons exposed to dust blowing off the dry bed of Old Wives Lake, Saskatchewan. Hefflin et al. (1994) found a small increase in emergency room visits for respiratory disorders on the day of a dust event (lag day 0) and 2 days after (lag day 2) in arid southeastern Washington State; however, a study conducted in the same region by Schwartz et al. (1999) found no increased mortality on dust storm days or the following day. Furthermore, Slaughter et al. (2005) found no consistent associations between any size fraction of PM and cardiac or respiratory emergency room (ER) visits or hospital admissions. In the adjacent state of Idaho, which has extensive agriculture and is prone to frequent dust events, Norton and Gunter (1999) found no correlation between PM10 (primarily mineral dust) levels and respiratory diseases in the general population. Additionally, Rublee et al. (2020) described a 4.8% increase in total ICU admissions on local dust storm days across the USA and 9.2% and 7.5% increases in respiratory admissions at lag days 0 and 5, respectively. These discrepancies among findings may be influenced by the capture of PM exposure information, short- or long-term exposure to high PM levels, the effects of finer and coarser particles on health, aerosol sources and compositions, weather patterns, land use, sociodemographic factors, spatiotemporal changes, regional or larger scale, and statistical analysis. Then, the potential variability in the health effects of aerosol exposures inspired us to look further into the health effects of a dust-prone environment in which dust events are increasing, and the information is scarce to inform the community.

In the southwest USA, Rodopoulou et al. (2014) found an association between emergency room visits for cardiovascular conditions and PM10 and PM2.5 concentrations during the warm season in Las Cruces, NM, USA, when the city regularly experiences high winds and dust episodes. Grineski et al. (2011) found increased odds of hospitalization for asthma and acute bronchitis among persons of all ages and adults on the day of dust events in the desert city of El Paso, Texas, and up to 3 days later, with stronger impacts among children (mostly for girls with acute bronchitis) and low-income adult groups. Herrera-Molina et al. (2021), who looked at associations between dust exposure and hospitalizations in El Paso County, Texas, found an increased relative risk of hospitalizations between 2010 and 2014 for multiple conditions associated with dust events at lags of 0 to 7 days after exposure.

The Southern High Plains (Llano Estacado) of southwest-central North America is semiarid in climate, flat in topography, subject to high winds, and prone to extended droughts (Howard et al. 2019). The years 2011–2012 were one of the hottest and driest periods on record in Texas, including the Llano Estacado, resulting in the region being a hotspot of dust events (Kandakji et al. 2021; Kelley and Ardon-Dryer 2021). In the Southern High Plains, the most highly populated city and the surrounding county of Lubbock, Texas (Online Resource 1, Figure SI) (county population 278,831 in 2010: US Census Bureau 2019), have a long dust event record (Lee and Tchakerian 1995; Lee et al. 1993; Stout 2001). The dominant form of airborne particulate matter in Lubbock is mineral dust (Gill et al. 2009), primarily windblown soil from surrounding agricultural lands and scrublands (Kandakji et al. 2021). It is among the top 10 windiest cities in the USA, with an annual average wind speed of ~ 6 m s−1 (Texas Weather Guide 2016); dust concentrations in Lubbock show a positive correlation with wind speed (Stout 2001). The frequency of windblown dust events in the southwestern USA has been increasing (Tong et al. 2017) and should continue to rise as climate changes (Schweitzer et al. 2018). Therefore, exposure to dust events and their particulate matter may pose an increasing hazard to human health in Lubbock, Texas.

Dust events in the Llano Estacado are characterized by very high concentrations of coarse particles (> 2.5 μm) (PMcoarse) but also contain high concentrations of smaller particles (< 2.5 μm) (PM2.5), which can exceed the 24-h US National Ambient Air Quality Standard of 35 µg m−3, leading to acute exposure of fine particulate matter on those days (Ardon-Dryer et al. 2022; Kelley et al. 2020; Ardon-Dryer and Kelley 2022). Exposure to these environmental conditions has previously been associated with several respiratory conditions. A “haboob lung syndrome” was defined by Lubbock physicians, in which severe lung disease ensued several days after a major dust storm (Panikkath et al. 2013). Elmassry et al. (2020) indicated that Lubbock residents experience allergies, asthma, and respiratory tract infections throughout the year, and Cladosporium (a mold that can cause allergies and asthma in some people) increases during days with dust storm events.

Aims

The aim of this study is to investigate whether short-term dust exposure (DE) in Lubbock, Texas, is associated with significant increases in hospitalizations on the day of exposure and up to 7 days afterward from 2010 to 2014.

Methodology

Associations with dust exposures were investigated for hospitalizations in the county of Lubbock, Texas, over a 5-year period from 2010 to 2014 using all International Classification of Diseases, Ninth Revision (ICD-9) codes (aggregated together); individual ICD-9 category code groups; neurodegenerative diseases (ND), mental illness (MI), mental disorders (MD), births, complications associated with birth, genitourinary system diseases, excessive or frequent menstruation, circulatory system diseases, and each component of cerebrovascular infarction — stroke, ischemia, atherosclerosis, veins and lymphatics, and other forms of heart disease; respiratory system diseases and each component of acute respiratory, pneumonia and influenza, asthma, diseases of the respiratory system, septicemia, associated diseases (AD), and each component of AD (Online Resource 1, Table SI).

Data sources

Hospital admissions

Hospitalization and aftercare services data were obtained from the Texas Hospital Inpatient Research Data Files (RDF) from the Texas Department of State Health Services (TDSHS), Center for Health Statistics, Austin, Texas, for Lubbock County — a total of 160,072 hospitalization registrations (Online Resource 1, Table SII). The years 2010–2014 were chosen because they included periods of relatively high and low dust event frequency (see Fig. 2) (Kelley et al. 2020; Kandakji et al. 2021). Data included the following variables: date of admission, census block group of the patient, patient age, sex, and principal ICD-9 diagnostic code. The principal diagnostic code was preferred over other diagnostic codes because it better captures the exacerbations of disease as opposed to other diagnostics due to existing diseases (Centers for Medicare and Medicaid Services 1990).

PM and weather

Hourly average PM2.5 concentrations (µg m−3) were obtained from the Texas Commission on Environmental Quality (TCEQ) website for its single monitoring station (C325) located in Lubbock, Texas, from 2010 to 2014. Days with blowing dust (days with at least one observation of “BLDU-blowing dust,” excluding “PTCHY BLDU-patchy blowing dust”), hourly visibility (km), hourly wind speed (m s−1), hourly and daily average temperature (°C), and hourly and daily average relative humidity (%) were extracted from the Meteorological Aerodrome Reports (METARs) (coded weather observations) from the US National Weather Service for Lubbock International Airport for 2010–2014.

Approach for determining dust exposure

Because of the evidence of the chronic health effects of dust exposure (DE) (Meng and Lu 2007), two approaches were identified for analysis: (i) dust exposure criterion 1 (DE1), a total of 183 days with blowing dust events, as indicated by at least one daily observation of the BLDU code in the METAR present weather code, excluding observations of patchy blowing dust, and (ii) dust exposure criterion 2 (DE2) for days with maximum hourly PM2.5 > 35 µg m−3, maximum hourly wind speed (WS) ≥ 9 m s−1, and minimum visibility (VS) < 13 km (74 days) (shown in Fig. 1). We estimated the effects of the current day’s exposures (lag day 0) and for the following seven days (lag days 1–7) on hospitalizations.

Fig. 1
figure 1

Monthly and yearly distribution of dust exposure days (DE1 and DE2) in Lubbock, Texas, from 2010–2014

Full size image

DE2 was used as an air pollution threshold that has some characteristics of a dusty day and that can be easily recognized with the combination of PM2.5, WS, and visibility levels. Hourly maxima of PM2.5 concentrations and wind speed and hourly minima in visibility were used to define DE2 because PM concentrations in Lubbock often display short-term peaks of high concentration during windblown dust events that average approximately 2 h in length (Kelley et al. 2020; Kelley and Ardon-Dryer 2021). A WS of 9 m s−1 was chosen based on measurements of threshold velocity for wind erosion of dust-emitting soils of the Llano Estacado (Stout and Zobeck 1996; Zobeck and Van Pelt 2006). Even though almost all of these days are dusty days, it is possible to assume that some haze and smoke days could have been included in this category (Ardon-Dryer et al. 2023).

Population increase or decrease between 2010 and 2014 was obtained from the US Census statistical data for the county of Lubbock, Texas (US Census Bureau 2019) to remove nonenvironmental confounding elements (population change).

Statistical analyses

Relationships between dust exposure in two approaches and hospital admissions were investigated using time series analysis to describe the exposure–response relationship with the lagged effect in time. More precisely, we used a quasi-Poisson regression model, the regression model with an overdispersal Poisson family, with application of distributed lag non-linear models (DLNM) for time series data. The DLNM approach allows us to model the relationship between outcomes and environmental exposures and describe the effects that are delayed in time, by specifying a cross-basis matrix of exposure histories over the same lag period for time series data (Gasparrini et al. 2010). Statistical analyses were performed using the dlnm package in R (version 4.0.5) and Python; statistical significance was determined by p < 0.05.

Model

A generalized linear model with quasi-Poisson or Poisson family was generated to determine associations between exposure (DE1 and DE2) and hospitalizations due to neurodegenerative diseases (ND); mental illness (MI); mental disorders (MD); births; complications associated with birth; genitourinary system diseases; circulatory system diseases; respiratory system diseases; septicemia; associated diseases (AD) and its components; and all ICD-9 categories and their components during an 8-day period (the day of the DE and each of the 7 days after exposure). Cross-basis matrices for maximum PM2.5, minimum visibility, maximum wind speed, daily average temperature, and daily average humidity were included in the regression model to account for the lagged effects of the predictors. Natural splines were used as a smoothing function of time for any time-dependent outcome predictors or confounders with long-term trends and seasonal patterns not explicitly included in the model (Touloumi et al. 2004). Long-term trends and seasonal patterns (time with 7 degrees of freedom/year and day of the year) were analyzed with a natural cubic regression spline. The full model has the following formula:

$$\begin{array}{c}Log\;E\left[Y\tau\right]=\alpha+\beta_1{\;{PM}_{2.5}}_{\tau,\tau+7}+\beta_2{\;VS}_{\tau,\tau+7}+\beta_3\;{WS}_{\tau,\tau+7}{+\gamma_1\;Exposure+\gamma}_2{Temp}_{\tau,\tau+3}{+\gamma}_3{\;Humd}_{\tau,\tau+5}+\\\gamma_4s(time)+\gamma_5s\left(day\;of\;year\right)+\gamma_6\;season+\gamma_7\;holiday+\gamma_8\;Weekday+\gamma_9\;Weekend+\gamma_{10}\;Population\end{array}$$

where E[Y \(\tau\)] is the expected value of the Poisson distributed variable, Y \(\tau\), indicating the daily admission count on a day \(\tau\) for each diagnosis with Var(Y \(\tau\)); \(\alpha\) refers to the intercept of the hospital admissions; \({\beta }_{i}\) refers to the coefficients associated with the PM2.5 concentration, visibility, and wind speed (i = 1, 2, 3); \({\gamma }_{j}\) the coefficients associated with other covariates (j = 1, 2, ⋯, 10); \({P{M}_{2.5}}_{\tau ,\tau +7}\) is the cross-basis matrices of PM2.5 maximum hourly level on day \(\tau\) with lags up to 7 days; \({VS}_{\tau ,\tau +7}\) is the cross-basis matrices of minimum visibility hourly level on day \(\tau\) with lags up to seven days; \({WS}_{\tau ,\tau +7}\) is the cross-basis matrices of maximum hourly wind speed on day \(\tau\) with lags up to seven days; Exposure is each dummy variable for DE1 or DE2 days; \({Temp}_{\tau ,\tau +3}\) is the cross-basis matrices of average temperature by two lag strata, 0 and 1–3; \({Humd}_{\tau ,\tau +5}\) is the cross-basis matrices of average relative humidity on day \(\tau\) with lags up to 5 days, assuming the effects of relative humidity are delayed up to five days; season is the indicator variable for cold (October to March) and warm (April to September) seasons; holiday is indicator of a holiday; Weekday is the days from Monday to Sunday; Weekend is a dummy variable for weekend (Saturday and Sunday); Population is the estimated population increase or decrease; and s(∙) denotes the smooth function of the variable. Several models were run for each outcome of interest, that is, neurodegenerative disease, mental illness, mental disease, births, complications associated with birth, genitourinary system disease, circulatory system disease, respiratory system disease, septicemia, AD (independently and aggregated), and ICD-9 categories (independently and aggregated).

The stepwise variable selection technique was performed to select the best model from the full model for each condition. In the model selection, PM, wind speed, minimum visibility, and exposure (DE1 or DE2) components were always included in the model by default, and a model with the subset of predictors was chosen that minimized the Akaike information criterion (AIC) by comparing AICs for inclusion and removal of the predictor. The relative risk (RR) for each associated diagnosis was calculated (e.g., RR per 10 μg m−3 increase in day-to-day maximum hourly PM2.5, 5-km decrease in day-to-day minimum hourly visibility, or 4.5 m s−1 (10 mph) increase in day-to-day maximum hourly wind speed) based on the final selected regression model. β1 represents the log of expected counts of hospital admissions per 10 μg m−3 change in PM2.5, β2 represents the log of expected counts of hospital admissions per 5 km decrease in visibility, and β3 represents the log of expected counts of hospital admissions per 4.5 m s−1 increase in wind speed, which can be estimated from the quasi-Poisson regression analysis.

Results

During the study period, there were a total of 160,072 hospitalizations (Online Resource 1, Table SII). Daily average admissions for all causes were 87.7 ± 24.8 with a minimum of 1 and a maximum of 152 hospitalizations per day. The greatest daily average admission was for births, with 10.2 ± 4.3 with a minimum of 0 and a maximum of 36 hospitalizations. The fewest admissions were for diseases of atherosclerosis, with only 0.2 ± 0.5 daily average hospital admissions from 2010 to 2014. The descriptive analysis of the specific condition-related hospitalizations is shown in Table 1.

Table 1 Count of total hospitalizations in Lubbock for 2010–2014 by each diagnosis potentially associated with DE
Full size table

The estimated average daily maximum of hourly PM2.5 concentration for the study period was 16.7 ± 18.6 μg m−3, and the minimum and maximum values were 2.9 μg m−3 and 288.6 μg m−3, respectively. The estimated average daily maximum of hourly wind speed for the study period was 9.3 ± 3.1 m s−1, and the minimum and maximum values were 3.1 m s−1 and 23.2 m s−1, respectively. The minimum visibility during the study period was 0.2 km, and the maximum was 10 miles (16.1 km), which is the maximum possible visibility in USA METAR: the daily average of hourly visibility was 11.6 ± 5.4 km (7.2 ± 3.4 miles). The average maximum daily temperature for the study period was 24.0 ± 10 °C, and the minimum and maximum values were − 10.0 and 43.0 °C, respectively. The descriptive analysis of these independent variables is shown in Table 2.

Table 2 Descriptive analysis of maximum hourly wind speed (m s−1), maximum hourly PM2.5 (μg m−3), minimum visibility (km), maximum and average temperature (°C), maximum relative humidity (%). Lubbock, TX, 2010–2014
Full size table

Summary of significant associations

The diagnoses (individually and as a group) with no significant results were mental illness (depression and anxiety), births, complications associated with births, genitourinary system (kidney and bladder cancer, renal failure, infections, and prostate disorder), menstruation, circulatory system (other forms of heart disease and veins and lymphatics (embolism)), and septicemia. The category codes from the ICD-9 with no significant results were 1 (infectious and parasitic diseases), 3 (endocrine, nutritional, and metabolic diseases, and immunity disorders), 5 (mental disorders (depressed mood, dementia)), 6 (nervous system and sense organs), 7 (circulatory system), 8 (respiratory system), 9 (digestive system), 10 (genitourinary system), 11 (complications of pregnancy, childbirth, and the puerperium), 12 (diseases of the skin and subcutaneous tissue), 14 (congenital anomalies), 15 (perinatal period), 16 (symptoms, signs, and ill-defined conditions), 17 (injury and poisoning), and 18 (aftercare services or therapies).

The results showing significant associations with dust exposure defined by the specified daily exposure to blowing dust or specified maximum hourly PM2.5 and maximum wind speed with a decrease in minimum visibility in terms of relative risks (RRs) and 95% confidence intervals (CIs) are illustrated in Fig. 2 (online resource, table SIV). An RR value > 1 indicates that individuals are more likely to be hospitalized from exposure to that variable. Significant associations were analyzed with the two dust exposure approaches (DE1 and DE2).

Fig. 2
figure 2

Risk ratios after dust exposure with DE1 and DE2 analysis are defined by a daily 10 µg m−3 increase in maximum hourly PM2.5, a 4.5 m s−1 increase in maximum wind speed, and a 5-km decrease in visibility with their respective lag days

Full size image

Associations with increased PM2.5

Significant associations for a 10 μg m−3 increase in daily maximum of hourly PM2.5 were found for all analytical approaches (unless stated otherwise) for the following conditions (see Fig. 2): atherosclerosis was associated on lag day 1 (except for DE2); asthma on lag day 0 (except for DE2); and acute respiratory disease on lag day 4; hematologic diseases were associated on lag day 4 (except for DE1); and ischemia disease was associated on lag day 0 (except for DE1); diseases of the musculoskeletal system and connective tissue on lag day 1; respiratory system disease on lag days 3, 4 and 5; and associated diseases (AD) on lag day 5.

Associations with increased wind speed

Significant associations were found for a 4.5 m s−1 increase in maximum hourly wind speed for all analytical approaches for the following conditions (see Fig. 2): neurodegenerative diseases were associated on lag day 0; renal disorder due to hypertension was associated on lag day 5; mental disorders were associated on lag days 1 and 6; and atherosclerosis was associated on lag day 2.

Associations with decreased visibility

Significant associations were found for a decrease of 5 km (3 miles) of minimum hourly visibility for all analytical approaches (unless stated otherwise) for the following conditions (see Fig. 2): acute respiratory disease was associated on lag day 0; renal disorders due to hypertension were associated on lag days 3 and 4; asthma was associated on lag day 6; cerebrovascular infarction or stroke was associated on lag day 5; neoplasms were associated on lag day 6; and associated diseases (AD) were associated on lag day 6. Mental disorders were associated on lag day 5 with DE2.

Comparisons of hospital admissions by gender and age

Since gender and age amplify health risks from air pollution (Collins et al. 2011), we examined how age and gender modulate the health effects of dust exposure in Lubbock County. Online Resource 1 and Table SIII show the age and gender distribution of hospitalized patients during the study period. Chi-squared tests were used to explore associations between hospital admission rates of each cause by subjects’ gender and age (under 1 year, 1–17 years, 18–64 years, and 65 years and over).

Hematologic diseases was the only one showing a higher female hospitalization (1.1%) than males (0.9%). Hospitalizations for respiratory-related diseases were higher for males than females (7.2% for male hospital admissions were due to respiratory system disease, compared to 5.8% for females; 0.7% of male hospitalizations vs. 0.5% of female hospitalizations were due to acute respiratory disease). However, there was no significant difference in hospital admissions by sex for asthma.

Age was found to be a significant factor associated with hospital admissions for the disease groups. Hospitalization rates were higher among older people for most diseases, while the younger age group (1–17) showed higher admissions for asthma (9.3%) and mental disorders (4.6%) than the other age groups.

Associations with weekend, season, and holidays

Hospitalizations that increased on weekdays compared to Sundays were due to all diagnoses associated with each analysis approach except for asthma (Online Resource 1, Table SV). Hospitalizations that decreased on holidays compared to nonholidays were due to ischemia (lag day 0), diseases of the musculoskeletal system and connective tissue (lag day 1), neurodegenerative diseases (lag day 0), mental disorders (lag day 1 and 6), atherosclerosis (lag day 2) (associated with a 4.5 m s−1 day-to-day increase in maximum hourly wind speed), circulatory system disease (lag day 0), neoplasms (lag day 0 and lag day 6) (associated with a 4.5 m s−1 increase in maximum hourly wind speed and a 5 km decrease in minimum visibility, respectively), and associated diseases (lag days 5 and 6) were associated with a 10 μg m−3 increase in maximum PM2.5 and a 5 km decrease in minimum visibility, respectively. Hospitalizations that decreased in the cold season compared to the hot season (p < 0.01) were due to asthma (lag day 0 and lag day 6) (associated with a 10 μg m−3 increase in maximum PM2.5 and a 5 km decrease in minimum visibility, respectively) and mental disorders (lag days 1 and 6).

Discussion

Small but significant increases in relative risks for hospitalization for multiple conditions on the day of or within a 7-day period after exposure to a dust event using two approaches were found for Lubbock County, Texas, for the period 2010–2014. Although associations were investigated for lags of 0 to 7 days, unlike results for El Paso, Texas (Herrera-Molina et al. 2021), no significant associations were found for any condition for lag 7. Some of the associations were modulated by season, weekday/weekend, and holiday effects. The significant associations and lag days were largely the same within each variable (PM2.5, wind speed, and visibility) for all two definitions of exposure (DE1 and DE2). The difference in associations is mostly between each variable of study. The diagnostic associations found in PM2.5 are different from those found in wind speed (except for atherosclerosis). The analysis with PM2.5 indicates more associations with respiratory system diseases; wind speed analysis shows more associations with neurodegenerative and mental disorders; visibility analysis shows a mix of diagnostic associations found in PM2.5 and wind speed. This could be because visibility is reflected by both PM2.5 and wind speed factors.

The potential mechanisms for the adverse effects of dust particles and exposure to PM2.5 are believed to generally originate from an inflammatory pathway (including neuroinflammation) and oxidative stress after inhalation (Hahad et al. 2020), which has been confirmed by in vitro (Costa et al. 2020) and in vivo (Hajipour et al. 2020) studies. Laboratory animals exposed to dust storm aerosols showed severe damage to the respiratory system followed by disruption of blood–brain barrier integrity, increased brain edema, inflammatory cytokine excretion, and oxidative stress in brain tissue (Hajipour et al. 2020). Dust may also be ingested, and the exposure of the gut to particulate matter initiates inflammatory responses within the small and large intestine and alters the gastrointestinal microbiome, which could trigger and accelerate the development of inflammatory disease (Kish et al. 2013). All these mechanisms may link dust exposure to cause diverse diseases.

Amounts of time after human exposure to particulate matter are required to cause new or additional inflammation and its effects to develop (Santos et al. 2008; Scapellato and Lotti 2007); thus, it may take several days (depending on the magnitude of exposure, the body system affected, and comorbidities) for sufficiently severe symptoms to manifest, causing the exposed person to seek hospitalization. In addition, different diseases associated with particulate matter exposure in different locations may have clinical manifestations with different severities, affecting time lags of medical attendance (Zhang et al. 2021). For example, in Lubbock, associated diseases (an aggregation of the most common causes of hospitalization) were associated with all models for dust exposure with a 10 µg m−3 increment of hourly PM2.5 at lag 5 (0.4% increase in risk (95% CI, 0.1–0.7%)), perhaps indicating a subtle effect of PM-triggered inflammation only manifested in the aggregate after 5 days.

Associations with hospitalizations for respiratory, cardiovascular, and cerebrovascular infarction diseases

Much previous research on the health effects of dust events has focused on respiratory and circulatory (cardiovascular and cerebrovascular) diseases, with dozens of investigations summarized in Aghababaeian et al. (2021), Lwin et al. (2023), Sadeghimoghaddam et al. (2021), and Zhang et al. (2016). In Lubbock, (nonacute) respiratory system disease hospitalizations were significantly associated with dust exposure-related PM2.5 increments on lag days 3, 4, and 5 (RR increase by 0.9% (95% CI, 0–1.8%), 1.2% (95% CI, 0.3–2.1%), and 1.1% (95% CI, 0.2–2.1%) per 10 μg m−3 increase in maximum hourly PM2.5, respectively), similar to observations made in dusty regions of China (Meng and Lu 2007; Tao et al. 2012) and the Canary Islands (López-Villarrubia et al. 2021). Acute respiratory disease, associated with mineral dust exposure in other studies (e.g., Geravandi et al. 2017; Watanabe et al. 2021), was associated with an increase in PM2.5 in Lubbock on lag day 4 (2.8% (95% CI, 0.1–5.6%) per 10 μg m−3 increase in maximum hourly PM2.5) and with a decrease in visibility at lag day 0 (8.0% (95% CI, 1.4–15.1%) per 5 km decrement in visibility). Associated with dust exposure-related PM2.5, hospitalization for ischemia was significantly increased at lag day 0 (2.8% (95% CI, 0.5–5.2%) per 10 μg m−3 increase in maximum hourly PM2.5); Dominguez-Rodriguez et al.’s (2021) systematic meta-analysis on the effects of desert dust found the incidence of cardiovascular mortality to be statistically significant at lag day 0. Cerebrovascular infarction hospitalization, which has been associated with Asian dust health effects in multiple studies (summarized in Hashizume et al. 2020), was associated in Lubbock on lag day 1 (RR increase by 2.0% (95% CI, 0–3.9%) per 10 μg m−3 increase in maximum hourly PM2.5) and on lag day 5 (RR increase by 2.4% (95% CI, 0.1–4.8%) for DE1 and 4.8% (95% CI, 1.6–8.2%) for DE2 per 5 km decrement in visibility).

Asthma has been widely associated with desert dust exposure worldwide (summarized, for example, in Zhang et al. 2016): a potential mechanism of dust asthma induction was described by Schweitzer et al. (2018). In Lubbock, we found an asthma–dust association with PM2.5 increments at lag day 0 (RR increase by 3.0% (95% CI, 0.1–5.9%) per 10 μg m−3 increase in maximum hourly PM2.5). Grineski et al. (2011) found that dust at lag day 0 had the largest odds ratios for the prediction of asthma hospitalization in El Paso, another dusty Texas city. In addition, Herrera-Molina et al. (2021) found an asthma–dust association, within the AD group, with PM10 increments at lag day 0, 1, and 2 (RR increase by 0.5% for day 0 (95% CI, 0.1% and 0.9%) per 100 μg m−3 increase of maximum hourly PM10 at lag day 0) in El Paso. Similar to the findings from Kanatani et al. (2010) in Japan and many follow-up studies globally, we found an increased hospitalization risk for dust-associated asthma in Lubbock for both children and adults.

Associations with hospitalizations for neurodegenerative diseases and mental disorders

The highest risk ratios overall in this study were for hospitalizations for neurodegenerative diseases (ND) associated with dust exposure at lag day 0 (25.0% (95% CI, 8.8–43.7%) for DE1 and 22% (95% CI, 7–39.1%) for DE2 per 4.5 m s−1 increase in max wind speed). Mental disorders were associated with wind speed at lag day 1 (5.1% (95% CI, 0.2–10.3%) per 4.5 m s−1 increase in max wind speed) and lag day 6 (6.7% (95% CI, 1.7–12%) per 4.5 m s−1 increase in max wind speed) and visibility at lag day 5 (2.2% (95% CI, 0.1–4.4%) per 5 km decrement in visibility). Fine particulate matter has recently been identified as a risk factor for dementia since it may affect cognitive function by neuroinflammation because of systemic inflammation or oxidative stress (Zhang et al. 2023). It has also been suggested that desert dust containing cyanotoxins from biological soil crusts is linked to ND (Sini et al. 2021); this appears unlikely in Lubbock, where dust is derived from agriculturally managed, uncrusted soils. Although a recent study indicates that PM2.5 from agriculture emissions shows greater rates of incident dementia (HR, 1.13; 95%CI, 1.01–1.27) and that PM2.5 sources of traffic, coal combustion, and wildfires presented the strongest associations (Zhang et al. 2023), except dust, it may be inherent to agricultural and road traffic activities as indicated by Arslan and Aybek (2012).,

The associations between PM2.5 exposure and ND and mental disorders have been presented in several recent reviews (Costa et al. 2020; Cristaldi et al. 2022; Fu et al. 2019; Hahad et al. 2020; Hajipour et al. 2020; Rauf et al. 2022). Lee et al. (2019a) suggested that the oxidative stress and neuroinflammation associated with inhaled PM2.5 are associated with an increased incidence of mental disease, showing that exposure to a 10 μg m−3 increment of PM2.5 was associated with significant increases in emergency admissions for mental disease at lag days 0–1. Lee et al. (2019b) found that exposure to Asian dust storms in Korea was associated with a significant increase in suicide risk on the day of dust storm occurrence. Bakian et al. (2015) found an association between PM2.5 exposure and suicide at lag day 2 in Salt Lake County, Utah (USA), and Yackerson et al. (2011) reported a slight but significant association between desert wind direction changes and suicide attempts and psychotic attacks in Israel.

Cristaldi et al. (2022) wrote, “exposure to PM2.5 promotes neuroinflammation processes, because through breathing the particles can reach the nasal epithelial mucosa and transferred to the brain through the olfactory bulb. Furthermore, exposure to PM2.5 has been associated with an increased expression of markers of neurodegenerative diseases (e.g., alpha-synuclein or beta-amyloid), which can contribute to the etiopathogenesis of neurodegenerative diseases.” Although Cristaldi et al. (2022) focused on PM2.5 exposure from diesel exhaust and traffic-related emissions, not soil particles, the association with wind and dust in Lubbock is somewhat novel. In vivo studies by Hajipour et al. (2020) showed that exposure to dust storm particles disrupted the integrity of the blood–brain barrier, increased brain edema, and increased inflammatory cytokine excretion and oxidative stress in brain tissue, inducing memory impairment. We echo Cristaldi et al.’s (2022) conclusion that “the potential cellular and molecular mechanisms of PM2.5 leading to neurodegenerative disease remain not entirely clear, and further studies need to be carried out on the topic.”

Associations with hospitalizations for other conditions

Diseases of the musculoskeletal system and connective tissue were associated with dust exposure in Lubbock through an increase in PM2.5 at lag day 1 (1.3% (95% CI, 0.2–2.4%) per 10 μg m−3 increase in maximum hourly PM2.5); hematologic diseases were associated with lag day 4 (2.8% (95% CI, 0.9–4.7%) per 10 μg m−3 increase in maximum hourly PM2.5); and atherosclerosis was associated with lag day 1 (4.4% (95% CI, 0.3–8.7%) per 10 μg m−3 increase in maximum hourly PM2.5) and with wind speed increments at lag day 2 (13.2% (95% CI, 0.5–27.5%) per 4.5 m s−1 increase in max wind speed). Hospitalization for neoplasms was associated with visibility on lag day 6 (3% (95% CI, 0.2–5.9%) per 5 km decrement in visibility). Hospitalizations for renal disorder due to hypertension were associated with a wind speed increment at lag day 5 (9.9% (95% CI, 1.3–19.4%) per 4.5 m s−1 increase in maximum wind speed) and with a decrease in visibility at lag day 3 (6.2% (95% CI, 1.5–11.2%) per 5 km decrement in visibility) and lag day 4 (7.4% (95% CI, 2.6–12.5%) per 5 km decrement in visibility).

Studies have found modest increases in musculoskeletal and connective tissue disease across China with short-term exposure to PM2.5 (Gu et al. 2020), as well as to immune and inflammatory responses, which were associated with exacerbations of autoimmune diseases due to exposure to PM (Zhao et al. 2019). These effects may be generally related to the oxidative and inflammatory properties of fine particulate matter (Zhao et al. 2019), including exposure to dust particles (Hajipour et al. 2020).

The role of location and composition

In addition to natural mineral matter from soil particles, a component of PM2.5 in Lubbock and elsewhere consists of organic compounds, including biological materials from bacteria and fungi (Elmassry et al. 2021), ultrafine particles (Ardon-Dryer et al. 2022), crystalline silica, and metals (Gill et al. 2009), all of which may have adverse human health effects. Pesticides and herbicides from agricultural lands of the Llano Estacado are dispersed into the wind (Peterson et al. 2020) and have been found within the Lubbock urban area (Miersma et al. 2003). These constituents could all play various roles in the specific health effects associated with dust exposure at Lubbock. Other dusty cities will have different meteorological conditions, different soil chemistries or biota, different local industries contributing PM, and different regional agricultural cropping systems, resulting in different physical, chemical, and biological compositions of their dust, which manifest different human health effects from dust exposure through different pathophysiological pathways of exposure to different particulate mixtures of materials. Thus, the diseases associated with dust in Lubbock may be different from those associated with dust exposure elsewhere.

For example, the associations and temporal lags indicated for Lubbock were moderately different than those found for the same time period for El Paso, Texas, although the two urban areas located in Texas are 550 km away from each other, which means they may experience different dust events and magnitudes. Herrera-Molina et al. (2021) used PM10, not PM2.5 or observed weather, as a basis of comparison to health effects in El Paso and did not consider visibility in the analysis. El Paso is a much larger city located within a much larger metropolitan area compared to Lubbock, with much more diverse industry. El Paso and adjacent Ciudad Juarez, Mexico, are surrounded by the largely undeveloped Chihuahuan Desert (Baddock et al. 2011) as opposed to the intensively agriculturally managed Llano Estacado surrounding Lubbock. In addition, there is a high percentage of dust-producing unpaved roads within the El Paso/Juarez metropolitan area (Kavouras et al. 2016); thus, the composition of El Paso and Lubbock dusts would be different, which can indicate different human health effects.

Limitations and opportunities for future research

Certain limitations must be considered regarding the datasets used and the extent of applicability of this study. We investigated the relationship of dust exposure to hospitalizations, rather than mortality (less frequent than hospitalization) or emergency room visitation or visits to healthcare providers (more frequent than hospitalization), which could be markers of greater or lesser effects of dust exposures on health, respectively. Only the principal diagnosis was obtained and considered for each patient; thus, the effects of comorbidities may have been overlooked. In addition, persons affected by some conditions, such as mental or neurogenerative diseases, may avoid seeking treatment altogether due to social stigmas (Shrivastava et al. 2012) or consult with their physician or other health professional rather than seek hospital treatment (Bushnell et al. 2005; Tylee and Walters 2007). Sensitivity analyses incorporating different parameters should be further investigated to examine the robustness of the parameter estimates across the DLNMs for different causes of hospitalization. Future studies should continue to investigate the relationship between exposure and health outcomes using data on an individual basis rather than census block-based information, which might yield more precise estimates of relative risk. In addition, subgroup analyses stratified by age groups, gender, and race could be conducted to account for the relative changes in hospitalization visits by age and gender.

There are many, potentially overlapping, sources and combinations of data to indicate dust exposure; we used two different approaches (DE1 and DE2), but there are others (Lwin et al. 2023). The lack of standardized protocols to identify and quantify dust exposure could result in inconsistent findings observed across health effects studies (Lwin et al. 2023). In this study, for one surrogate of dust exposure (DE1), we used days with one present weather code to represent dust (BLDU). However, records of dusty conditions, whether made by human weather observers or automated meteorological sensors, may be mischaracterized as “haze” or other phenomena or omitted, causing inaccuracies (Ardon-Dryer et al. 2023). In addition, due to the sometimes-ephemeral nature of dust events in Lubbock (Ardon-Dryer and Kelley 2022), a dust event observed at the airport may not occur in some parts of Lubbock, and some dust events exposing city residents may not have impacted the airport.

As part of the dust exposure criterion DE2, this investigation used data on PM2.5 rather than PM10 or PMcoarse, even though dust events are dominated by coarse particles and PM10 would be a preferable metric of comparison since the TCEQ air monitoring station in Lubbock only measured PM2.5 concentrations. Since there is only one air monitoring station in Lubbock, it is not certain that the high PM2.5 levels recorded there would represent similar conditions in other parts of the city or vice versa. Hourly maxima rather than daily averages of PM2.5 were used since high-resolution aerosol monitoring shows that Lubbock dust events are characterized by short-term “spikes” of extremely high concentrations superimposed on a background of lower particulate matter levels (Ardon-Dryer and Kelley 2022; Ardon-Dryer et al. 2022). This maximum hourly value would proxy acute exposure or overexposure to aerosols, but some diseases might be associated with chronic, lower-intensity aerosol exposures over long periods of time better represented by daily or even monthly or annual averages of PM. Additional research could explore the effects of short-term exposures at lags longer than 7 days and investigate longer-term exposures to dust in Lubbock and other dusty cities.

Although we were able to identify a pattern of association between neurodegenerative diseases, mental disorders (bipolar, schizophrenia, and unspecified episodic mood disorder and unspecified psychosis), cerebrovascular infarction (cerebral infarction, cerebral ischemia, etc.), and dust exposure, we did not find an association with mental illness (depression and anxiety). Several other recent studies have found significant increases in hospitalizations for mental illness associated with increased concentrations of PM10 and coarse particles (Gao et al. 2017; Qiu et al. 2019) as well as PM2.5 (Lee et al. 2019a). Individuals experiencing symptoms consistent with neurodegenerative or mental conditions, especially anxiety and depression, often have their disorder go unrecognized because they do not explain their psychological symptoms explicitly and because they commonly consult their primary physician and/or mental health provider instead of going to the hospital (Bushnell et al. 2005; Tylee and Walters 2007). In addition, ~ 50% of people with depression never consult a healthcare provider, 95% never enter secondary mental health services, and many more are unrecognized and untreated (NICE 2009). Consultations with mental health providers, rather than hospitalizations, may more accurately capture mental health associations with dust exposure.

Conclusions, implications, and recommendations

Hospitalizations due to neurodegenerative diseases, atherosclerosis, renal disorders due to hypertension, acute respiratory disease, mental disorders, neoplasms, asthma, hematologic diseases, cerebrovascular infarction (stroke), ischemia, diseases of the musculoskeletal system and connective tissue, diseases of the respiratory system, and associated diseases were significantly positively associated with dust exposure in Lubbock, Texas, at different lag periods (numbers of days) after exposure, indicated from higher to lower risk. The significant associations at different lag days were largely similar regardless of the form of exposure (DE1, DE2). Associations with AD (hospital admissions for all causes each associated with at least 5% of hospitalizations) could be due to small effects that emerge from aggregated hospitalizations but not individual disease categories for individual diagnoses yet still indicate a short-term association of hospitalizations, broadly considered, with dust exposure in Lubbock.

Diagnoses significantly positively associated with dust exposure increased on weekdays compared to Sunday (p < 0.01), suggesting that patients avoided hospitalization on weekends: persons with symptoms that they perceive as mild may delay presentation to a hospital during weekends (Bachner and Zuba 2022; Rotstein et al. 1997). Hospitalizations that decreased on holidays compared to nonholidays (p < 0.01) were due to ischemia, musculoskeletal system and connective tissue diseases, respiratory system disease, neurodegenerative diseases, atherosclerosis, circulatory system disease, neoplasms, and associated diseases. Similar to weekends, patients may avoid seeking hospital care on holidays (Castner et al. 2016; Rotstein et al. 1997).

The results show that dust-associated hospitalizations due to asthma increase in Lubbock during the warm season; a systematic review of the effect of PM2.5 on asthma made the same finding (Fan et al. 2016), pointing out the combined effects of heat, increased pollen, and other air pollutants such as ozone increase during the warm season. Hospitalizations due to mental diseases also increased during the warm season, an association that has been found with PM2.5 elsewhere (Lee et al. 2019a). Hematologic diseases were the only condition that had a significantly higher hospitalization rate for females than for males. Hospital admissions for asthma and mental disorders were highest for children (1–17 years old); most other conditions showed higher hospitalization rates for older persons. Further studies will be needed to explore how associations between hospitalization for diseases and PM, wind speed, and visibility are adjusted by season and subject-specific age and gender.

In the North American Great Plains, climate and land use change are increasing dust loadings (Lambert et al. 2020), and climate change is expected to greatly increase the public health consequences of dust in Southwest North America (Achakulwisut et al. 2018). Globally, more than 40% of the world’s population lives in drylands at present; that percentage is expected to increase as drylands expand with climate change (Spinoni et al. 2021), and water stresses are also projected to increase on global drylands (Stringer et al. 2021), which are likely to also increase dust loadings and human health consequences for cities such as Lubbock.

Future research should investigate the association of dust with emergency department and health professional visits, as well as mortality, to complete a broader examination of dust exposure associations with health in the Lubbock area and to examine potential similarities or differences to more limited prior studies in other North American dusty agricultural zones (e.g., Hefflin et al. 1994; Schwartz et al. 1999). It is also advisable to consider ways to capture information about acute PM exposures occurring over periods of time of only minutes, as occur in dust storms (Ardon-Dryer and Kelley 2022), and long-term exposures (months or years), in order to evaluate any immediate and chronic health effects they may induce or those of long-term exposure. Similar investigations should be pursued in dusty cities where long-term records of both PM2.5 and PM10 data are available to determine the potential differential effects of finer and coarser particles on health indicators. Finally, equivalent studies should be pursued in dryland agricultural cities in other regions, where weather patterns, agricultural practices, soils, and thus aerosol sources and compositions would be different, to coax out potential compositional, geographic, and land use variability in the health effects of aerosol exposures.

These findings have important implications because the majority (~ 92%) of the world’s population resides in areas where particulate matter concentrations are greater than the World Health Organization (WHO) guidelines and approximately 2/5 of the world’s population resides in drylands (WHO 2016); therefore, the association between dust events and disease cannot be ignored. Approximately half of all dryland inhabitants on Earth are poor, more than a billion people in total, and have habitually been neglected in development processes; this “forgotten billion” living where dust events are most prevalent is particularly vulnerable to their adverse effects (Middleton and Kang 2017).

Public policies and individual actions are essential to reduce the human health effects of dust events. Examples include physical wind erosion control measures on agricultural lands (Nordstrom and Hotta 2004), improving forecasting of dust storms (Campbell et al. 2022), improving air quality alerts and their communication to the public (Gladson et al. 2022), and increasing public education and outreach on air pollution and its human health effects (Rice et al. 2021). Implementing such actions will reduce the adverse consequences of exposure to dust for residents of Lubbock and other dusty cities.