Skip to main content

Carbon monoxide poisoning: a prediction model using meteorological factors and air pollutant

Abstract

Background

While the influence of meteorology on carbon monoxide (CO) poisoning has been reported, few data are available on the association between air pollutants and the prediction of CO poisoning. Our objective is to explore meteorological and pollutant patterns associated with CO poisoning and to establish a predictive model.

Results

CO poisoning was found to be significantly associated with meteorological and pollutant patterns: low temperatures, low wind speeds, low air concentrations of sulfur dioxide (SO2) and ozone (O38h), and high daily temperature changes and ambient CO (r absolute value range: 0.079 to 0.232, all P values < 0.01). Based on the above factors, a predictive model was established: “logitPj = aj - 0.193 * temperature - 0.228 * wind speed + 0.221 * 24 h temperature change + 1.25 * CO - 0.0176 * SO2 + 0.0008 *O38h; j = 1, 2, 3, 4; a1 = -4.12, a2 = -2.93, a3 = -1.98, a4 = -0.92.” The proposed prediction model based on combined factors showed better predictive capacity than a model using only meteorological factors as a predictor.

Conclusion

Low temperatures, wind speed, and SO2 and high daily temperature changes, O38h, and CO are related to CO poisoning. Using both meteorological and pollutant factors as predictors could help facilitate the prevention of CO poisoning.

Background

Carbon monoxide (CO) poisoning is the main cause of emergency department admissions, and it is a leading cause of death from poisoning worldwide [1]. Studies showed that CO poisoning affects 50,000 people per year in the United States [2] and that most cases are unintentional [3]. Domicile-related CO poisoning mostly occurs in night and is difficult to detect [4]. Hence, the development of effective prediction for CO poisoning is of great public health importance. In China, domicile-related CO poisoning prevails with the main sources of CO being biomass and coal stoves [5].

Many studies have explored how meteorological patterns influence the prevalence of CO poisoning, which largely occurs in winter and is associated with low meteorological temperatures, low barometric pressure and low wind speeds [6]. Weather patterns influence pollutant distributions, and their combination might thus play important roles in the initiation of domicile-related CO poisoning [7]. As local real-time meteorological and pollutant data are easy to obtain, they may serve as effective predictors. However, few data are available on pollutant patterns related to CO poisoning. Over the past two decades, several studies have been published on many aspects of CO poisoning in the realms of pathophysiology, diagnosis, and clinical management along with evidence-based recommendations for optimal clinical practice [4]. Though many improvements have been made, CO poisoning remains a serious problem worldwide and especially in developing countries [8]. Developing effective forecasting methods could help better prevent the poorly understood and collectively appearing results of poisoning, but their prediction has been largely not explored and their effects remain unclear.

Publications on CO poisoning and associated deaths occurring in mainland China are very limited [8, 9], and there is a paucity of work on how meteorological and pollutant patterns influence CO poisoning. Hence, the aim of the present study was to characterize meteorological and pollutant patterns associated with CO poisoning in Liuzhou (a city in southern China) and to explore a prediction model for this issue.

Results

Monthly distribution of CO poisoning

The cases of CO poisoning showed a pattern of annual cyclicity with a peak occurring from December to March (Fig. 1) and with cases ranging from 0 to 500 per month. The cases of CO poisoning in January of 2016 and 2017 was at top, but an abruptly decrease occurred in that month of 2015 year. Cases of CO poisoning recorded from December to March account for more than 80% of all cases occurring from 2015 to 2017.

Fig. 1
figure1

Monthly distribution of CO poisoning from 2015 to 2017 in Liuzhou

Missing value processing and air pollutant levels

The days with missing value of pollutant factor (approximately 0.2–0.5%) were excluded from the analysis. From 2015 to 2017, the daily mean concentrations of SO2, NO2, CO, and O38h were 24.82 μg/m3, 26.11 μg/m3, 1.048 mg/m3, and 87.51 μg/m3, respectively, and daily mean concentrations of PM10 and PM2.5 were 74.98 and 53.13 μg/m3, respectively (Additional file 1: Table S1). Overall, air pollutant concentrations were low, though some were in a moderate level according to China Ambient Air Quality Standards (GB 3095–2012).

Correlations between meteorological and pollutant parameters and the occurrence of CO poisoning

Relationships between the occurrence of CO poisoning and meteorological parameters (temperature, air pressure, humidity, wind direction and wind speed) and pollutant parameters (SO2, NO2, PM10, PM2.5, CO, and O38h) were analyzed. We found that CO poisoning was significantly and inversely associated with temperatures, 24-h atmospheric pressure changes, wind speeds and ozone (O38h) (r ranges: − 0.162 to − 0.409, all Ps < 0.001) and positively associated with 24-h temperature changes, atmospheric pressure, humidity, and pollutants (SO2, NO2, PM2.5, PM10, and CO) (r ranges: 0.085 to 0.371, all Ps < 0.01) (Table 1 and Additional file 1: Table S2). Overall, the association of temperature-related indices with CO poisoning was stronger among meteorology indices, while it was stronger for NO2 (followed by CO and PM2.5) among pollutant indices.

Table 1 Pearson correlation between CO poisoning admission and meteorological parameters

A partial correlation analysis showed that temperature, wind speed, SO2 and O38h (R ranges: − 0.079 to − 0.152, all Ps < 0.01) were inversely correlated with CO poisoning incidence, and 24-h temperature change and air CO concentration (R ranges: 0.098–0.232, all Ps < 0.01) were positively correlated with CO poisoning incidence (Table 2). No significant correlations were found for other indices. Thus, in further prediction analysis, these six indices with significant partial correlations with CO poisoning were used for predictors.

Table 2 Partial correlation between CO poisoning admission and meteorological and environmental factors

Logistic regression analyses of daily (Fig. 2) and monthly (Figure S1) CO poisoning levels were conducted separately, and the result showed that among the all parameters evaluated, CO poisoning was most closely associated with temperature, followed by air pressure and O38h; among environmental factors, CO poisoning was most closely associated with CO concentration, followed by PM10.

Fig. 2
figure2

Logistic regression of CO poisoning by daily mean meteorological and pollutant factors. Abbreviation: CO, carbon monoxide; O3, ozone; SO2, sulfur dioxide

Ordinal logistic regression of CO poisoning risk levels and prediction model establishment

Results of our ordinal regression analyses showed that temperature (β: -0.193, OR: 0.82, 95% CI: 0.80, 0.85) was inversely associated with CO poisoning, whereas 24-h temperature difference (β: 0.221, OR: 1.25, 95% CI: 1.18, 1.31) and CO (β: 1.25, OR: 3.48, 95% CI: 2.82, 4.14) (all Ps < 0.001) were positively associated with CO poisoning (Table 3). Wind speeds (β: -0.228, OR: 0.80, 95% CI: 0.54, 1.06; P = 0.086) and SO2 (β: -0.018, OR: 0.98, 95% CI: 0.96, 1.00; P = 0.058) were marginally associated with CO poisoning. Null associations were found for O38h (β: 0.0008, OR: 1.00, 95% CI: 1.00, 1.01; P = 0.766). These results showed that CO poisoning was closely related with the ambient CO concentration, being 3.48 times more likely to occur along with per 1 mg/m3 increase in ambient CO concentrations and 1.25 times more likely to occur at per 1 °C increase in 24 h temperature change. In addition, we found that 0.82 times less possibility of CO poisoning occurred along with per 1 °C decrease in temperature.

Table 3 Ordinal logistic regression of CO poisoning admission by meteorological and environmental factors in total patients

Based on the regression coefficient, prediction models for CO poisoning grades risk was established as follows: 1: using single meteorological factor as prediction factors and 2: using combined meteorological and pollutant factor as prediction factors. Single-factor and combined-factor prediction was conducted separately to fit a better model. A lower Akaike information criterion (AIC) value denotes a stronger model fit. Two fitting models were shown below:

  1. 1:

    logitPj = aj - 0.195 * temperature - 0.356 * wind speed + 0.224 * 24 h temperature change

$$ {\displaystyle \begin{array}{l}\mathrm{j}=1,2,3,4\\ {}\mathrm{a}1=-5.32,\mathrm{a}2=-4.15,\mathrm{a}3=-3.23,\mathrm{a}4=-2.20;\kern0.5em \mathrm{AIC}=1777\end{array}} $$
(1)
  1. 2:

    logitPj = aj - 0.193 * temperature - 0.228 * wind speed + 0.221 * 24 h temperature change + 1.25 * CO - 0.018 * SO2 + 0.0008 *O38h;

$$ {\displaystyle \begin{array}{l}\mathrm{j}=1,2,3,4;\\ {}\mathrm{a}1=-4.12,\mathrm{a}2=-2.93,\mathrm{a}3=-1.98,\mathrm{a}4=-0.92;\kern0.5em \mathrm{AIC}=1767.\end{array}} $$
(2)

According to AIC standards, we identified that the combined-factor model with pollutant factors had a better goodness of fit than the single meteorological factor model.

Examination of the CO poisoning prediction model

Prediction models were examined using actual CO poisoning levels in 2017 and the prediction potential were analyzed, which were shown in Fig. 3 and Additional file 1: Table S3. The boxplots showed that the median actual cases of CO poisoning gradually increased with predictive grades (P-diff < 0.001), i.e. the actual poisoning case number was lowest in grade 1 and increased gradually in grade 2–5 (Fig. 3). To assess the comprehensive prediction power of the models, we calculated average effective prediction rate in two extreme grades (Grade 1 & 5). Single-factor model showed an effective rate of 83.8% (average of 145/152 [Grade 1] and 47/65 [Grade 5]), slightly lower than that of 84.4% (average of 142/152 [Grade 1] and 49/65 [Grade 5]) in the combined-factor model. In addition, single-factor model predicted 0 days for Grade 4 and was presumed to be a little weaker in predicting serious poisoning grades compared with the combined model. Altogether, the single meteorological factors can predict the occurrence of CO poisoning cases, but the combined model with pollutant factors is more accurate and performs better especially for serious grades of CO poisoning.

Fig. 3
figure3

Median comparison of CO poisoning cases of different forecast grade. a Combing meteorological and environmental factors as prediction factor, b single meteorological factors as prediction factor

Discussion

Our data show that for the city of Liuzhou, meteorological and pollutant patterns of low temperatures, low wind speeds, low SO2 and high CO and O38h levels were related to CO poisoning incidence. Among these, a prominent positive association between air CO and poisoning grades and an inverse association between temperature and poisoning grades were found. A prediction model based on those parameters can predict poisoning cases well, and the combined model with air pollutant factor are more accurate and perform better than models based on meteorological factors alone. Considering both meteorological and pollutant parameters as prediction factors will thus play a role in the prevention of CO poisoning.

Some previous studies have reported that domicile-related CO poisoning mainly occurred in cold months and at lower temperatures. Climate conditions influence air pollutant distributions. Cases of CO poisoning was severe worldwide. However, few studies have investigated the association between CO poisoning and climatic factors such as wind speed, daily temperatures and air pollutants [6, 10]. A population-based case-control study of Taiwan found a 2.15-fold increase in CO poisoning mortality in cold periods with maximum daily temperatures of less than 18.4 °C than in warm periods with daily maximum temperatures of 27.1 °C or greater [10]. The onset of colder weather typically results in behavior changes (e.g., higher levels of indoor heating). Xie J et al. found that wind speed was the main factor that affects CO poisoning accidents which occurred in dwellings and that ventilation, smoke emissions, differences between indoor and outdoor temperatures, low daily gas pressure changes could aggravate CO poisoning occurrence [11]. In addition, low wind velocities, ground-level inversion and fog aggravate domicile-derived CO poisoning, as they impede the rise of combustion emissions through chimneys and the dispersal of indoor CO [7, 12]. CO poisoning outbreaks are also related to storms and power outages [13].

We consistently found temperatures to be the most important protection against CO poisoning along with wind speed. Climate conditions were also found to be closely related to air pollutants. Low temperatures and wind speeds lead to limited air movements, potentially increasing concentrations of air pollutants. A series of studies have shown that ambient air pollutants SO2, CO, O3, PM10, PM2.5, and NO2 can induce a variety of acute symptoms [14] and are related to hospitalizations for respiratory [15, 16] and cardiovascular diseases [17, 18]. Notably, CO is reportedly strongly associated with mortality and heart failure [19]. While few studies have explored the influence of air pollutants on CO poisoning, except one study found a poor  pollutant distribution on days with poisoning events [7]. Epidemiological studies have also shown that air pollutants contribute to daily mortality [20] and that the association remains even at very low pollution levels [21]. The specific effects of air CO and other pollutants on the incidence of CO poisoning are unknown. Our data show strong independent positive effects of pollutant (SO2, CO and O38h) distributions and CO poisoning. In particular, a 3.48-fold higher risk of CO poisoning was found for per 1 mg/m3 increase in air CO concentration (OR [95% CI], 3.48 [2.82, 4.14], P < 0.001). Among underlying mechanisms involved, it was assumptive that high outdoor CO concentrations might impede the diffusion of indoor CO gas. To our knowledge, no other study has investigated the correlation between ambient CO concentration and indoor CO poisoning. Future studies must clarify underlying mechanisms involved.

The exposure dose of CO depends on its air concentration, the exposure time and the patient’s breathing volume per minute. Indoor smoke elevates baseline carboxyhemoglobin levels [22, 23]. Other indoor sources include exhaust from poorly ventilated gas- and oil-burning home appliances and the infiltration of outdoor pollutants [24]. In unpolluted ambient air, concentrations of CO range from 0.02–1.0 ppm; however, in urban areas, CO concentrations can increase 10-fold in periods of atmospheric stagnation [25, 26] (i.e., during temperature inversions in winter or when sedentary air masses form in summer). CO levels of approximately 100 ppm can be detected in kitchens when a gas stove is used for cooking or when there is a gas leak [24]. The Chinese government recommends that the indoor CO concentration not exceed a maximum of 50 ppm. Hence, preventing CO leaks and reducing the ambient CO concentration are central to the prevention of CO poisoning.

A single factor may not account for the comprehensive causes of CO poisoning. Previous studies have examined the effects of meteorological change on CO poisoning, and our data show that pollutant-related factors might also play an important role in CO poisoning incidence. Meteorological and pollutant patterns—low air temperatures, wind speeds, and SO2 levels and high 24-h temperature differences, CO levels and O38h concentrations—might increase the incidence of domicile-related CO poisoning. In this study, we established two prediction models: a meteorological model covering temperature, 24 h temperature changes, and wind speed and a model covering meteorological and pollutant-related factors (SO2, CO and O38h). Both models effectively predict CO poisoning occurrence. The combined model is superior to the single factor model in the following respects. First, the combined model is more accurate when applied to extreme grades (84.4%, which is slightly higher than the value of 83.8% found for the single-factor model). Second, the combined-factor model is more effective at predicting serious poisoning grades than the single-factor model. Third, the single-factor model cannot identify cases of level 4 poisoning. These results show that both meteorological and air pollutant parameters must be considered for the prediction of CO poisoning [12]. Future studies must validate these findings and explore more underlying mechanisms through which outdoor environmental conditions influence CO poisoning. In addition, high-risk populations (e.g., the elderly and children) must be closely considered.

CO poisoning is preventable and avoidable. Automobiles and cooking and heating appliances have long been regarded as the main environmental sources of CO [27]. Gas stoves and water heaters are very popular in the city of Guangxi. A gas leak due to improper installation or aging parts can result in CO poisoning. Thus, safety education designed to increase awareness of CO poisoning and of ways to reduce CO production should be effective at reducing the occurrence of domicile-related poisoning [28, 29]. Public service warnings and government supervision can also help prevent poisoning events, especially during storms [30]. As we found that combined-factor prediction shows strong prevention potential, incorporating CO poisoning predictions into weather broadcast programs could raise public awareness of CO poisoning occurrence [7]. Effective air purification measures taken by environmental protection agencies could also improve the prevention of CO poisoning, and installing CO detectors in residences have favorable cost benefits [3], which can be used as an important secondary prevention measure [31]. Finally, prompt treatment measures including fast and correct diagnosis and effective therapy should decrease the incidence of death from CO poisoning [30].

This study presents some strengths. First, we are the first to analyze the influence of pollutants on CO poisoning and to establish a prediction model of CO poisoning based on meteorological factors alone and on a combination with environmental parameters. Second, in applying our models to actual cases of CO poisoning, we found that the model considering both meteorological and pollutant parameters can predict CO poisoning well and performs better than the model based on meteorological factors alone in its accuracy and predictive potential for high poisoning grades. This study also has some limitations. First, the medical records considered in this study represent cases of CO poisoning in the studied region but not include cases with missed diagnoses, patients who died close to sources of CO poisoning and were not admitted to a designated hospital. Nevertheless, cases of CO poisoning considered in this study were collected from four designated hospitals for CO poisoning, which ensured a more comprehensive and reliable dataset. Second, while we established a prediction model based on multiple meteorological and environmental factors and validated the model in several ways, the model only xaccounts for part etiological causes of CO poisoning.

Conclusions

Temperatures, wind speeds, daily temperature changes, ambient SO2, CO and O38h serve as significant predictors of CO poisoning. The proposed prediction models based on meteorological and pollutant parameters can help prevent CO poisoning by targeting high-risk populations. The developed model considering both meteorological and pollutant parameters offers superior predictive power than the proposed single-factor model, especially for days of high poisoning risk. Future studies must validate our findings.

Methods

Population and data source

CO poisoning data for 2015–2017 were collected from the emergency department of four hospitals which were designated by Liuzhou Health Commission for CO poisoning. The medical records of all eligible patients were reviewed, and cases of suicide and attempted suicide and accidental CO poisoning were excluded. Meteorological and pollutant data for the same period were obtained from the Liuzhou Meteorological Bureau and Liuzhou Environmental Protection Agency, respectively. All meteorological and pollutant parameters were taken from the 6 representative monitoring sites.

CO poisoning gradation

To construct a prediction model of CO poisoning, we graded CO poisoning into 5 levels according to the number of daily cases of CO poisoning as follows: Level 1, 0 cases of CO poisoning daily; Level 2, 1 case of CO poisoning daily; Level 3, a moderate risk of daily CO poisoning with > 1 and ≤ 3 cases; Level 4, a high risk of daily CO poisoning with > 3 and ≤ 7 cases; and Level 5, a very high risk of daily CO poisoning with > 7 cases.

Data and statistical analysis

First, to investigate possible correlations between the occurrence of CO poisoning and meteorological and pollutant parameters, records of meteorological (temperature, gas pressure and daily changes, humidity, wind direction, and wind speed) and air pollutant parameters (sulfur dioxide (SO2), nitrogen dioxide (NO2), inhalable particulate matter (PM)10, PM2.5, CO, and ozone (O38h)) for Liuzhou for 2015–2017 were collected. Mean daily meteorological and pollutant parameters and cases of CO poisoning were analyzed.

Further, Pearson correlation analyses were performed to examine the correlation between meteorological and pollutant factors and CO poisoning. Partial correlation analyses were conducted with related parameters controlled. Indices of meteorological and pollutant parameters showing a significant partial correlation (P < 0.05) with CO poisoning were selected as prediction factors. Logistic regression was used to evaluate the association between CO poisoning levels and daily and monthly mean values of the single prediction factors. Multivariable ordinal logistic regression was used to evaluate coefficients and odds ratios (95% CI) per 1 unit increase in the predictive factors based on data for 2015 and 2016. Then, the coefficient of predictors obtained was used to establish a prediction model of CO poisoning.

We used data for 2017 to examine the predictive power of both models. From the probability model established, we obtained predictive CO poisoning risk levels for each day of 2017 based on intraday meteorological and pollutant information and in turn determined the most likely risk levels. The Kolmogorov-Smirnov test was used to examine differences in average cases of CO poisoning based on pairwise prediction grades. All analyses were conducted using R3.5 (https://www.r-project.org/) where a 2-tailed probability of less than 0.05 was considered statistically significant.

Availability of data and materials

The data access committee comprises the first author and the corresponding author, and there is no restriction to data access. De-identified individual-level data are available upon reasonable request to H.L.R at drruanhailin@163.com.

Abbreviations

CO:

Carbon monoxide

O3 :

Ozone

PM10 :

Particle matter 10

PM2.5 :

Particle matter 2.5

NO2 :

Nitrogen dioxide

SO2 :

Sulfur dioxide

References

  1. 1.

    Weaver LK, Hopkins RO, Chan KJ, Churchill S, Elliott CG, Clemmer TP, Orme JF Jr, Thomas FO, Morris AH. Hyperbaric oxygen for acute carbon monoxide poisoning. N Engl J Med. 2002;347:1057–67.

    CAS  Article  Google Scholar 

  2. 2.

    Hampson NB, Weaver LK. Carbon monoxide poisoning: a new incidence for an old disease. Undersea Hyperb Med. 2007;34:163–8.

    CAS  PubMed  Google Scholar 

  3. 3.

    Ran T, Nurmagambetov T, Sircar K. Economic implications of unintentional carbon monoxide poisoning in the United States and the cost and benefit of CO detectors. Am J Emerg Med. 2018;36:414–9.

    Article  Google Scholar 

  4. 4.

    Rose JJ, Wang L, Xu Q, McTiernan CF, Shiva S, Tejero J, Gladwin MT. Carbon monoxide poisoning: pathogenesis, management, and future directions of therapy. Am J Respir Crit Care Med. 2017;195:596–606.

    CAS  Article  Google Scholar 

  5. 5.

    Lu Q, Lv W, Tian J, Zhang L, Zhu B. Risk factors for non-occupational carbon monoxide poisoning: Anshan prefecture, Liaoning province, China, 2011-2012. PLoS One. 2015;10:e0129121.

    Article  Google Scholar 

  6. 6.

    Du T, Zhang Y, Wu JS, Wang H, Ji X, Xu T, Li Y, Xu L, Lewin MR. Domicile-related carbon monoxide poisoning in cold months and its relation with climatic factors. Am J Emerg Med. 2010;28:928–32.

    Article  Google Scholar 

  7. 7.

    Gijsenbergh FP, Vispoel M, Poppe H, Delooz HH. Weather influence on the prevalence of carbon monoxide intoxications. Acta Clin Belg Suppl. 1990;13:96–7.

    CAS  PubMed  Google Scholar 

  8. 8.

    Chen F, Ye Y, Wei Q, Chen J, Wu H, Yan Y, Liao L. Non-fire related carbon monoxide poisoning in Sichuan, China: a 9-year study (2008-2016). Iran J Public Health. 2019;48:458–64.

    PubMed  PubMed Central  Google Scholar 

  9. 9.

    Li F, Chan HC, Liu S, Jia H, Li H, Hu Y, Wang Z, Huang W. Carbon monoxide poisoning as a cause of death in Wuhan, China: a retrospective six-year epidemiological study (2009-2014). Forensic Sci Int. 2015;253:112–8.

    CAS  Article  Google Scholar 

  10. 10.

    Shie HG, Li CY. Population-based case-control study of risk factors for unintentional mortality from carbon monoxide poisoning in Taiwan. Inhal Toxicol. 2007;19:905–12.

    CAS  Article  Google Scholar 

  11. 11.

    Xie J, LI L, Guo W, Ying S, Xuan C. Meteorological factors of carbon monoxide poisoning accident in vernacular dwelling and its numerical simulation. J Meteorol Environ. 2010;26:63–9.

    Google Scholar 

  12. 12.

    Fu G, Li Y, You F. Analysis on the meteorological conditions of gas poisoning event. Chin Meteorol Mon. 2008;34:103–9.

    Google Scholar 

  13. 13.

    Hampson NB, Stock AL. Storm-related carbon monoxide poisoning: lessons learned from recent epidemics. Undersea Hyperb Med. 2006;33:257–63.

    CAS  PubMed  Google Scholar 

  14. 14.

    Wiwatanadate P. Acute air pollution-related symptoms among residents in Chiang Mai, Thailand. J Environ Health. 2014;76:76–84.

    CAS  PubMed  Google Scholar 

  15. 15.

    Tian L, Qiu H, Pun VC, Lin H, Ge E, Chan JC, Louie PK, Ho KF, Yu IT. Ambient carbon monoxide associated with reduced risk of hospital admissions for respiratory tract infections. Am J Respir Crit Care Med. 2013;188:1240–5.

    Article  Google Scholar 

  16. 16.

    Nhung NTT, Schindler C, Dien TM, Probst-Hensch N, Kunzli N. Association of ambient air pollution with lengths of hospital stay for Hanoi children with acute lower-respiratory infection, 2007-2016. Environ Pollut. 2019;247:752–62.

    Article  Google Scholar 

  17. 17.

    Chang CC, Tsai SS, Ho SC, Yang CY. Air pollution and hospital admissions for cardiovascular disease in Taipei, Taiwan. Environ Res. 2005;98:114–9.

    CAS  Article  Google Scholar 

  18. 18.

    Dennekamp M, Akram M, Abramson MJ, Tonkin A, Sim MR, Fridman M, Erbas B. Outdoor air pollution as a trigger for out-of-hospital cardiac arrests. Epidemiology. 2010;21:494–500.

    Article  Google Scholar 

  19. 19.

    Morris RD, Naumova EN, Munasinghe RL. Ambient air pollution and hospitalization for congestive heart failure among elderly people in seven large US cities. Am J Public Health. 1995;85:1361–5.

    CAS  Article  Google Scholar 

  20. 20.

    Moolgavkar SH. Air pollution and daily mortality in three U.S. counties. Environ Health Perspect. 2000;108:777–84.

    CAS  Article  Google Scholar 

  21. 21.

    Vedal S, Brauer M, White R, Petkau J. Air pollution and daily mortality in a city with low levels of pollution. Environ Health Perspect. 2003;111:45–52.

    Article  Google Scholar 

  22. 22.

    Varon J, Marik PE, Fromm RE Jr, Gueler A. Carbon monoxide poisoning: a review for clinicians. J Emerg Med. 1999;17:87–93.

    CAS  Article  Google Scholar 

  23. 23.

    Scherer G. Carboxyhemoglobin and thiocyanate as biomarkers of exposure to carbon monoxide and hydrogen cyanide in tobacco smoke. Exp Toxicol Pathol. 2006;58:101–24.

    CAS  Article  Google Scholar 

  24. 24.

    Exposure guidelines for residential indoor air quality: a report of the Federal - Provincial Advisory Committee on Environmental and Occupational Health. Ottawa: Health Canada. 1995;Cat no H46–2/90-156E. Available: www.hc-sc.gc.ca/ehp/ehd/catalogue/bch_pubs/90ehd156.htm. Accessed 28 May 2002.

  25. 25.

    National ambient air quality objectives for carbon monoxide: desirable, acceptable and tolerable levels. Ottawa: Environment Canada and Health Canada. 1994. Available: http://publications.gc.ca/Collection/En42-17-8-1994E.pdf.

  26. 26.

    Campbell ME, Benson BA, Muir MA. Urban air quality and human health: a Toronto perspective. Can J Public Health. 1995;86:351–7.

    CAS  PubMed  Google Scholar 

  27. 27.

    Hawkins M. Carbon monoxide poisoning. Eur J Anaesthesiol. 1999;16:585–9.

    CAS  Article  Google Scholar 

  28. 28.

    Fruchtman Y, Perry ZH, Leibson T, Vered Cohen L, Lifshitz M, Leibovitz E. Pediatric carbon monoxide poisoning in southern Israel: a cross-sectional study. Pediatr Emerg Care. 2020;36(11):532–6.

    PubMed  Google Scholar 

  29. 29.

    Wirth I, Varchmin-schultheiss K, Schmeling A. Negligent homicide caused by exhaust gas escaping from a manipulated chimney. Arch Kriminol. 2011;227:164–73.

    PubMed  Google Scholar 

  30. 30.

    Hampson NB, Piantadosi CA, Thom SR, Weaver LK. Practice recommendations in the diagnosis, management, and prevention of carbon monoxide poisoning. Am J Respir Crit Care Med. 2012;186:1095–101.

    CAS  Article  Google Scholar 

  31. 31.

    Hampson NB, Weaver LK. Residential carbon monoxide alarm use: opportunities for poisoning prevention. J Environ Health. 2011;73:30–3.

    CAS  PubMed  Google Scholar 

Download references

Acknowledgments

We appreciate all team member for their contribution in the data collection and analysis and manuscript writing.

About this supplement

This article has been published as part of BMC Proceedings Volume 15 Supplement 1, 2021: Selected Articles from the 20th International Conference on Bioinformatics & Computational Biology (BIOCOMP 2019): proceedings. The full contents of the supplement are available online at https://bmcproc.biomedcentral.com/articles/supplements/volume-15-supplement-1.

Funding

Publication charges for this article have been jointly funded by Research Project of Health Department of Guangxi Zhuang Autonomous Region, China (No z2014395) and Guangxi Emergency and Medical Rescue Talent Base Guangxi University Key Laboratory of Emergency Medicine, China (No gxjz201416). The funders had no role in the study design, data collection and analysis, decision to publish, or writing of the manuscript.

Author information

Affiliations

Authors

Contributions

ZJH, HLR conceived and designed the research; HLR, WSD, YW, SSY and JBC, WLH collected the data; WSD, YW and SSY analyzed the data; ZJH, HLR draft the paper; ZJH and HLR co-edited, revised the paper; and HLR had primary responsibility for final content. All authors have read and approved the final manuscript.

Corresponding author

Correspondence to Zhuo-Jun Hu.

Ethics declarations

Ethics approval and consent to participate

The Medical Ethics Committees of Liuzhou Worker’s Hospital and other designated hospitals, Guangxi Liuzhou meteorological bureau, Guangxi Liuzhou environmental protection bureau approved this study, and all participants signed a consent form.

Consent for publication

All authors read and gave approval to publication of the final version of the manuscript.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Table S1.

Distribution of daily incidence counts and overall daily pollutant and meteorological measures. Table S2. Pearson correlation between CO poisoning admission and pollutant parameters. Figure S1. Logistic regression of CO poisoning by monthly mean meteorological and pollutant factors. Table S3. A comparison of actual CO poisoning level in 2017 and predictive CO poisoning grade based on the combination of meteorological and pollutant factors.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ruan, HL., Deng, WS., Wang, Y. et al. Carbon monoxide poisoning: a prediction model using meteorological factors and air pollutant. BMC Proc 15, 1 (2021). https://doi.org/10.1186/s12919-021-00206-7

Download citation

Keywords

  • CO poisoning
  • Meteorological factors
  • Air pollutant levels
  • Prediction