Relationship between Air Pollution and Outpatient Visits for Nonspecific Conjunctivitis

Abstract

Purpose.: Past studies present evidence of associations between air pollution and human ocular symptoms; however, to the knowledge of the authors, research investigating the hazardous effects of air pollution on nonspecific conjunctivitis is nonexistent. This study investigates the relationship between air pollution and outpatient visits for nonspecific conjunctivitis in Taiwan. A multiarea analysis was conducted to examine and assess the risks of short-term effects of particulate matter (PM), nitrogen dioxide (NO₂), sulfur dioxide (SO₂), ozone (O₃), and carbon monoxide on nonspecific conjunctivitis.

Methods.: Data were collected from outpatient visits for nonspecific conjunctivitis from seven air-quality–monitoring areas. To find immediate and lag effects of air pollution, an area-specific, case-crossover analysis was performed and a meta-analysis with random effects was used to combine the area-specific results.

Results.: The effects on outpatient visits for nonspecific conjunctivitis are strongest for O₃ and NO₂, with a 2.5% increase (95% confidence interval [CI], 0.9–4.1) for a 16.4 ppb (parts per billion) concentration rise in O₃ and a 2.3% increase (95% CI, 0.7–3.9) for an 11.47 ppb concentration rise in NO₂. Effects are also found for particulate matter with an aerodynamic diameter ≤ 10 μm (PM₁₀) and SO₂. Effects are more prominent in winter because the analysis was stratified according to season.

Conclusions.: The air pollutants NO₂, SO₂, O₃, and PM₁₀ increase the chances of outpatient visits for nonspecific conjunctivitis and have no evident lag effects.

Air pollution causes diverse health effects and diseases, including cardiovascular disease, respiratory tract problems, eye irritation, neurologic disorders, cancer, and death. ^{1 –3} Dense innervations in the ocular surface are extremely sensitive to environmental agents. ⁴ Furthermore, the eyes are protected by only a thin layer of tear film from potentially damaging exterior influences. ⁵ Thus, human eyes are very susceptible to ill effects of air pollution. 

The ill effects of air pollution on human eyes are mostly irritation and inflammation, with conjunctivitis being a significant problem. ^6,7 Many research investigations have attempted to discover the effects of environmental toxins on the ocular surface. One study has shown that, among persons traveling regularly through high-polluted areas, a considerable portion suffers from subclinical ocular surface changes. ⁸ Another has found that air pollution contributes to cell altering and ocular surface inflammation, leading to ocular discomfort. ⁹  

Conjunctivitis is one of the most commonly diagnosed conditions in ophthalmologic outpatient and emergency room visits. ^10,11 More than 40% of ophthalmologic outpatient visits are diagnosed with conjunctivitis annually, according to data from the Bureau of Taiwan National Health Insurance. Conjunctival disorders induced by air pollution can be subclinical ocular surface changes, ⁸ but on many occasions with serious discomforts such as burning and irritation require clinical visits. Furthermore, chronic exposure to air pollution can induce cellular change, such as goblet-cell hyperplasia in the ocular surface. ¹² The discomforts of ocular diseases manifested in burning, irritation, itching, and tearing interfere with people's daily activities including work efficiency and road safety. Treating ocular diseases by steroid eye drops occasionally results in cataracts, glaucoma, and other severe side effects, which can lead to permanent vision loss. ¹³  

Numerous studies have investigated the relationships between air quality and outpatient or emergency room visits regarding the respiratory or cardiovascular systems. However, only a few have examined the relationships for eye disease. To the best of our knowledge, this is the first large-scale study investigating the impact of air pollution on ocular health based on the data from ophthalmologic outpatient visits. This multiarea study assessed the risk of short-term effects of air pollution on nonspecific conjunctivitis in Taiwan between 2007 and 2009. The research focuses on the significance of air pollution impact on ocular health, and the results may help in establishing a warning system to monitor the components of air pollution that have higher ocular risk. 

Methods

Health Data

Data from ophthalmologic outpatient visits between 2007 and 2009 were obtained from the National Health Insurance Database of Taiwan. More than 99% of Taiwan's residents receive national health insurance, making this database a valuable medical research resource. Outpatient visits for nonspecific conjunctivitis were selected according to the International Classification of Diseases, 9th revision (ICD-9) diagnosis codes. The following codes were included: 372.00 (nonspecific acute conjunctivitis), 372.01 (serious conjunctivitis except viral infection), 372.10 (chronic conjunctivitis), 372.11 (simple chronic conjunctivitis), 372.20 (blepharoconjunctivitis), 372.30 (other undefined conjunctivitis), and 372.39 (other conjunctivitis). The original data were further filtered with statistical software and only the data from ophthalmology specialists were used for calculation. The following data were excluded: patients visited either the same or another ophthalmologist more than once within 10 days and the diagnosis of conjunctival disorders was changed. Only outpatient visits with the aforementioned ICD-9 codes as the major diagnosis were included in this study. 

Air Quality Data

The study used daily ambient air-quality data between 2007 and 2009 that were extracted from the database maintained by the Taiwan Environmental Protection Administration (EPA). Taiwan's air-quality–monitoring network divides Taiwan into seven air-quality control areas: North, Chu-Miao, Central, Yun-Chia-Nan, Kao-Ping, Yilan, and Hua-Tung. There are 70 fully automatic monitoring stations measuring air quality in these areas. Air pollutants enclosed in the analysis were particulate matter (PM₁₀ and PM_2.5), nitrogen dioxide (NO₂), sulfur dioxide (SO₂), ozone (O₃), and carbon monoxide (CO). CO was analyzed via the infrared method, NO₂ by the chemiluminescence method, SO₂ by the ultraviolet fluorescence method, O₃ by the ultraviolet absorption method, and PM₁₀ and PM_2.5 by the beta-ray attenuation method. Temperature, rainfall, and relative humidity data were concurrently obtained from the database. 

Analytical Strategy

This study used a multiarea case-crossover design to investigate the association between daily air pollution concentrations and outpatient visits for nonspecific conjunctivitis of the selected ICD-9 codes. Multicity case-crossover analysis had been used to investigate the short-term effects of air pollutants on the development of respiratory infections and wheezing. ^{14 –16} A case subject was regarded as a control person on days with no outpatient visits. This study selected the control days by matching the day of the visit with the day of the week in the same month and year. In addition, a case-crossover analysis was performed by matching every third day from the case day in the same month and year for a sensitivity analysis. In the analysis of every third day, the day-of-week variable was included in the regression model. Same-day mean temperature, rainfall, and relative humidity were used in the models to control the potential influence of weather conditions. 

Because the relationship between air pollution and nonspecific conjunctivitis may vary according to season, seasonal stratified analyses were also performed. The effects from the same day up to 5 prior days were calculated by computing the moving averages as averages of exposure lags. For example, the 2-day moving average (lag 0–1) was computed as the mean of the same day and previous days. The associations between air pollution and outpatient visits for nonspecific conjunctivitis were analyzed in each area separately. To estimate the average effect for all the areas, the area-specific results were combined using a random-effect(s) meta-analysis. ¹⁷ The results are expressed as percentage increases in each air-pollution component for an interquartile range (IQR) increase in exposure. The IQRs are the average of the seven areas. A conditional logistic regression was used to analyze the data (SPSS18, IBM Corp, USA; Stata 11, StataCorp LP, USA). 

Results

The seven air-quality–monitoring areas span the entire island of Taiwan. The North area, located in the northern tip of Taiwan, has the largest population (nearly 9 million) and the greatest number of air-quality–monitoring stations (18 stations). The Yilan area, which is located in the northeast of the island, has the least population (nearly 0.5 million) and the fewest air-quality–monitoring stations (two stations). Table 1 shows the daily mean and SD of outpatient visits for nonspecific conjunctivitis for an entire year and according to season. More than 20 million outpatient visits occurred during the time periods examined. The number of outpatient visits for nonspecific conjunctivitis is higher in winter compared with that in summer for all areas. 

Table 1.

 

Descriptive Statistics for Daily Counts of Outpatient Visits in Each Area, in Total and by Season (Mean ± SD)

Monitoring Area	All (×103)	Winter (×103)	Summer (×103)
North	8.90 ± 3.67	9.17 ± 3.78	8.58 ± 3.47
Chu-Miao	1.25 ± 0.50	1.31 ± 0.52	1.58 ± 0.47
Central	4.88 ± 1.90	5.11 ± 1.93	4.58 ± 1.74
Yun-Chia-Nan	3.84 ± 1.47	4.00 ± 1.54	3.58 ± 1.39
Kao-Ping	4.25 ± 1.64	4.43 ± 1.71	4.58 ± 1.55
Yilan	0.46 ± 0.22	0.48 ± 0.23	0.58 ± 0.21
Hua-Tung	0.43 ± 0.19	0.45 ± 0.20	0.58 ± 0.17

Table 2 lists the concentration of air pollution and weather conditions. The mean levels of air pollutants varied across these seven areas, and the degrees of variation were different among pollutants. The Kao-Ping and the North areas displayed the highest levels of air pollution, reflecting heavy traffic and dense industrial activities. 

Table 2.

 

Environmental Variables for Seven Monitoring Areas (Mean ± SD)

Variable	North	Chu-Miao	Central	Yun-Chia-Nan	Kao-Ping	Yilan	Hua-Tung
Temperature, °C
All	23.40 ± 3.50	23.20 ± 3.50	23.70 ± 3.40	24.40 ± 3.60	25.60 ± 3.10	23.50 ± 3.40	24.80 ± 3.00
Winter	19.60 ± 4.00	19.40 ± 4.10	20.20 ± 4.00	21.10 ± 3.80	22.90 ± 3.40	20.00 ± 3.60	22.10 ± 3.10
Summer	27.20 ± 3.40	27.00 ± 3.30	27.20 ± 2.70	27.60 ± 2.50	28.30 ± 2.00	27.00 ± 3.40	27.50 ± 2.70
Relative humidity, %
All	75.10 ± 6.10	73.90 ± 7.40	74.40 ± 6.10	76.00 ± 6.80	76.50 ± 7.80	76.00 ± 7.30	74.80 ± 7.30
Winter	74.50 ± 5.90	73.60 ± 6.00	72.30 ± 6.10	75.10 ± 6.40	71.80 ± 7.90	74.80 ± 7.50	70.10 ± 7.50
Summer	75.40 ± 7.20	74.20 ± 8.20	76.30 ± 6.60	76.80 ± 7.10	81.00 ± 7.50	78.80 ± 7.10	79.30 ± 7.30
Rainfall, mm/h
All	0.24 ± 0.72	0.21 ± 0.73	0.45 ± 2.00	0.90 ± 3.99	0.27 ± 1.26	0.43 ± 1.21	0.23 ± 0.73
Winter	0.15 ± 0.47	0.11 ± 0.40	0.16 ± 1.22	0.29 ± 4.38	0.04 ± 0.22	0.50 ± 1.33	0.15 ± 0.59
Summer	0.34 ± 0.90	0.32 ± 0.94	0.74 ± 2.52	1.21 ± 3.60	0.49 ± 1.74	0.36 ± 1.08	0.31 ± 0.83
CO, ppm
All	0.58 ± 0.21	0.41 ± 0.14	0.49 ± 0.17	0.42 ± 0.14	0.50 ± 0.19	0.40 ± 0.12	0.41 ± 0.12
Winter	0.60 ± 0.22	0.45 ± 0.15	0.57 ± 0.15	0.50 ± 0.12	0.63 ± 0.15	0.44 ± 0.13	0.46 ± 0.11
Summer	0.56 ± 0.19	0.37 ± 0.12	0.41 ± 0.15	0.34 ± 0.12	0.38 ± 0.14	0.37 ± 0.11	0.36 ± 0.10
NO2, ppb
All	21.00 ± 7.15	14.10 ± 5.51	17.20 ± 5.87	15.30 ± 5.58	19.70 ± 8.69	10.70 ± 3.10	8.86 ± 2.80
Winter	21.50 ± 7.61	15.60 ± 5.66	20.10 ± 5.11	19.00 ± 4.34	25.60 ± 7.32	11.70 ± 3.01	10.20 ± 2.39
Summer	20.50 ± 6.62	12.70 ± 4.94	14.40 ± 5.10	11.50 ± 3.93	13.90 ± 5.46	9.79 ± 2.89	7.51 ± 2.51
SO2, ppb
All	5.40 ± 2.45	3.36 ± 1.44	2.74 ± 1.23	3.99 ± 1.11	7.54 ± 3.09	2.77 ± 1.17	2.16 ± 0.70
Winter	4.87 ± 2.48	3.59 ± 1.72	2.87 ± 1.27	4.38 ± 1.18	8.84 ± 3.02	2.59 ± 1.27	2.33 ± 0.83
Summer	5.92 ± 2.30	3.13 ± 1.04	2.60 ± 1.18	3.60 ± 0.89	6.25 ± 2.58	2.96 ± 1.03	2.00 ± 0.47
O3, ppb
All	28.40 ± 10.02	28.90 ± 10.20	29.00 ± 10.30	34.40 ± 11.00	30.40 ± 12.40	27.30 ± 9.61	26.60 ± 9.30
Winter	28.00 ± 8.32	28.60 ± 8.70	27.30 ± 8.01	30.60 ± 9.06	31.80 ± 11.10	28.10 ± 7.81	28.30 ± 7.10
Summer	28.80 ± 11.50	29.40 ± 11.60	30.80 ± 11.80	32.20 ± 12.70	29.10 ± 13.50	26.60 ± 11.10	24.80 ± 10.78
PM10, μg/m3
All	52.80 ± 24.40	48.10 ± 23.80	58.20 ± 25.20	73.30 ± 36.90	79.80 ± 36.20	41.90 ± 20.10	34.30 ± 21.97
Winter	55.30 ± 27.40	53.10 ± 26.50	67.70 ± 25.40	92.80 ± 37.10	104.00 ± 29.40	42.60 ± 22.60	37.10 ± 25.34
Summer	50.30 ± 20.60	43.20 ± 19.60	48.80 ± 21.10	53.90 ± 24.30	55.60 ± 24.20	41.20 ± 17.30	31.50 ± 17.58
PM2.5, μg/m3
All	28.00 ± 15.30	23.80 ± 28.50	37.70 ± 17.90	40.80 ± 19.20	47.40 ± 22.80	20.10 ± 12.00	18.10 ± 8.78
Winter	29.50 ± 17.50	31.50 ± 17.40	43.80 ± 17.70	49.90 ± 18.00	62.00 ± 18.60	21.80 ± 13.00	19.10 ± 9.53
Summer	26.50 ± 12.50	25.60 ± 13.10	31.50 ± 15.70	31.90 ± 15.90	32.90 ± 16.60	18.40 ± 10.70	17.20 ± 7.86

Table 3 lists the P values for homogeneity and significant heterogeneity (at a significance level of 0.05) among the results of the seven air-quality areas, regarding NO₂ and SO₂. This signifies that the study results of the impact of air pollution on ocular health differ from area to area. For obtaining the average results of the seven areas, a meta-analysis was conducted in conjunction with random effect(s). 

Table 3.

 

Percentage Increase (95% CI) of Outpatient Visits for an IQR Increase in Air Pollution: Combined Results across Seven Area and P-Value for Homogeneity Test

Pollutant	Percentage (95% CI)	IQR	P-Value for Homogeneity
CO lag 0	0.2 (−0.4 to 0.8)	0.32	0.06
CO lag 0–1	0.3 (−0.9 to 1.5)	0.28	0.23
CO lag 0–2	0.5 (−1.3 to 2.3)	0.28	0.58
CO lag 0–3	0.4 (−1.5 to 2.3)	0.25	0.71
CO lag 0–4	0.6 (−1.9 to 3.1)	0.24	0.66
CO lag 0–5	0.9 (−2.3 to 4.1)	0.22	0.54
NO2 lag 0	2.3 (0.7 to 3.9)	11.47	0.11
NO2 lag 0–1	2.3 (0.4 to 4.2)	10.56	0.008
NO2 lag 0–2	2.4 (−0.1 to 4.9)	9.97	0.08
NO2 lag 0–3	2.8 (−0.4 to 6.0)	9.61	0.003
NO2 lag 0–4	2.5 (−1.0 to 6.0)	8.80	0.007
NO2 lag 0–5	2.3 (−1.3 to 5.9)	8.18	0.005
SO2 lag 0	1.8 (1.2 to 2.4)	4.60	0.02
SO2 lag 0–1	1.6 (0.8 to 2.4)	4.18	0.007
SO2 lag 0–2	1.8 (−0.3 to 3.9)	3.66	0.02
SO2 lag 0–3	2.0 (−0.3 to 4.3)	3.37	0.07
SO2 lag 0–4	2.2 (−0.6 to 5.0)	3.17	0.05
SO2 lag 0–5	2.7 (−0.2 to 5.2)	3.16	0.02
O3 lag 0	2.5 (0.9 to 4.1)	16.14	0.81
O3 lag 0–1	2.4 (0.3 to 4.5)	15.69	0.68
O3 lag 0–2	2.7 (−0.1 to 5.5)	14.09	0.41
O3 lag 0–3	2.6 (−0.4 to 5.6)	13.45	0.77
O3 lag 0–4	2.8 (−0.9 to 6.5)	12.78	0.74
O3 lag 0–5	3.2 (−1.1 to 7.5)	12.30	0.62
PM10 lag 0	0.9 (0.2 to 1.6)	44.05	0.08
PM10 lag 0–1	0.9 (0.1 to 1.7)	42.80	0.21
PM10 lag 0–2	1.0 (−0.2 to 2.2)	40.49	0.38
PM10 lag 0–3	1.1 (−0.4 to 2.6)	38.15	0.32
PM10 lag 0–4	1.1 (−0.6 to 2.8)	37.71	0.39
PM10 lag 0–5	1.3 (−0.5 to 3.1)	37.10	0.37
PM2.5 lag 0	−0.3 (−0.7 to 0.1)	27.00	0.13
PM2.5 lag 0–1	−0.5 (−1.1 to 0.1)	24.31	0.06
PM2.5 lag 0–2	−0.6 (−1.4 to 0.2)	21.81	0.12
PM2.5 lag 0–3	−1.0 (−2.3 to 0.3)	18.28	0.13
PM2.5 lag 0–4	−1.3 (−3.0 to 0.4)	17.56	0.15
PM2.5 lag 0–5	−1.5 (−3.8 to 0.8)	16.67	0.15

Table 3 also shows the combined results of the seven areas for a single day and the moving averages for nonspecific conjunctivitis. O₃ has the greatest impact with outpatient visits for nonspecific conjunctivitis, followed by NO₂ and SO₂. A weaker association is found for PM₁₀. No associations are found with CO and PM_2.5 for both single-day and moving averages. Most associations were found for lag 0 and lag 0–1, and the strongest for lag 0 with a 2.5% rise (95% confidence interval [CI], 0.9–4.1) for a 16.1 parts per billion (ppb) concentration increase in O₃. For all the pollutants, no significant associations were present for the averages of lags longer than lag 0–1. 

Results from the sensitivity analyses in which the control days were chosen from every third day from the case day in the same month and year show a pattern of associations (the lag structure and relative strength of associations across pollutants) similar to that of the main analysis; however, the strength of associations is weaker in the sensitivity analysis, despite a larger number of control days. 

When the data were stratified according to seasons, the associations for NO₂ and SO₂ were stronger in winter than those in summer, and the associations showed no difference between winter and summer for O₃ and PM₁₀ (Fig. 1). For NO₂ and SO₂, the associations for nonspecific conjunctivitis for the lag 0 in winter were found to be a 2.8% (95% CI, 0.2–5.4) and a 2.3% (95% CI, 0.3–4.3) increase for an IQR increase, and a 1.9% (95% CI, −0.8 to 4.6) and a 1.6% (95% CI, −0.1 to 3.3) increase for an IQR increase in summer. For O₃ and PM₁₀, the associations for nonspecific conjunctivitis for the lag 0 in winter were found to be a 2.6% (95% CI, 0.2–5.0) and a 2.5% (95% CI, 0.7–4.3) increase for an IQR increase, and a 1.2% (95% CI, 0.3–2.1) and a 1.0% (95% CI, 0.3–1.7) increase for an IQR increase in summer. 

Figure 1.

 

Results according to season for nonspecific conjunctivitis. Expressed as percentage increase (95% CI) in the risk of nonspecific conjunctivitis and IQR increase in air pollution.

Discussion

In the present study, an analysis was conducted for each of seven air-quality–monitoring areas situated in different geographical areas in Taiwan (north, center-north, central, and center-south, southwest, northeast, and southeast of Taiwan). These areas have different characteristics in climates, population density, business, and traffic conditions. As a result, ocular health effects may differ; therefore, analysis was conducted in each area separately. 

The research investigations exploring the relationship of air pollution and respiratory disorders have shown that air pollution can induce and worsen respiratory disorders. ^18,19 Previous studies show that approximately 90% of rhinitis cases experience ocular symptoms at least 1 day a week, and some researchers consider conjunctival mucosa and respiratory mucosa react similarly to exogenous stimuli. ^20,21 Contrasting the present study with and comparing research results with those of the respiratory system may yield significant parallels and differences. Respiratory morbidity has been shown to associate with groups of air pollutants, including two or more of CO, NO₂, SO₂, O₃, and PM. ^{22 –24} Many of the present study's results exhibit patterns in agreement with the investigations of respiratory disorders, whereas significant contrasts are also apparent. 

A study in Copenhagen found a link between incident wheezing symptoms in infants and air pollution (PM₁₀, NO₂, CO) with a 3- to 4-day lag. ²⁵ Associations between airway symptoms and air pollution appeared to increase or persist at longer lags (up to 5 days). ²⁴ Some researchers, examining the link between air pollution and emergency room visits for asthma in children and adults, ²⁶ asserted that risk value increased as longer lags were included in the moving averages. No similar lag effect was found in this study, however. This is reasonable because air pollution affecting the eyes mostly results in irritation, ^6,7 which should be felt immediately once air pollutants have come into contact with the eyes. 

Most studies show that PM significantly affects the respiratory system. ^22,23,27 In this study, however, only weak effects were found for PM₁₀, and no association was found for PM_2.5. This is possibly due to the effect of tears cleansing the ocular surface. The pollutants of PM that reach the ocular surface do not cause a change in pH as do other aerosol pollutants. PM can be cleared out of the ocular surface in a short time, ²⁸ whereas in the respiratory system, PM penetrates deep and stays in the smaller airway. 

Villeneuve and colleagues ²⁶ examined the link between air pollution and emergency room visits for asthma, showing that NO₂ and CO have especially pronounced associations. Their results also show that CO has a strong association in warmer seasons, and the estimated risks increased with lengthier lags are embedded in moving averages. However, in the present study, CO had only an insignificant influence on nonspecific conjunctivitis outpatient visits. This may be explained by the absence of ocular irritation behavior of CO. ²⁹  

Bourcier and colleagues ⁶ investigated short-term associations between ophthalmologic emergency room visits and air pollution. They concluded a relation between NO₂, O₃, PM₁₀, and SO₂ and conjunctivitis, with which our findings of this research are in complete agreement, even to the details of the lag effects. 

The mechanisms regarding the effects of air pollutants on conjunctivitis are not well known. Changes in the lacrimal pH, caused by the acidification of tears in an atmosphere with a high-oxidant power (NO₂, SO₂), could irritate the ocular surface. ³⁰ O₃ can cause severe irritation of respiratory tract mucosa, and the effects are also apparent on ocular mucosa. In this study, only PM with a larger diameter (PM₁₀) had a significant impact on the number of outpatient visits. This might be due to coarse PM, causing stronger foreign body sensations, whereas fine PM is easily cleaned from the ocular surface with tears, without inducing ocular discomfort. 

This study, which combined and integrated air pollution and ophthalmologic data to investigate associations between outpatient visits for nonspecific conjunctivitis and air pollution levels found that the air pollutants NO₂, SO₂, O₃, and PM₁₀ can increase chances of outpatient visits, suggesting possible causes for nonspecific conjunctivitis. The study shows that a favorable air-pollution–monitoring infrastructure, combined with established interdisciplinary research relationships can provide continued examination of mechanisms and changes in the eyes related to air pollution. 

Footnotes

 Disclosure: C.-J. Chang, None; H.-H. Yang, None; C.-A. Chang, None; H.-Y. Tsai, None