Article Text

Download PDFPDF
Association between temperature changes and uveitis onset in mainland China
  1. Handan Tan1,
  2. Su Pan1,
  3. Zhenyu Zhong1,
  4. Jing Shi1,
  5. Weiting Liao1,
  6. Guannan Su1,
  7. Aize Kijlstra2,
  8. Peizeng Yang1
  1. 1 The First Affiliated Hospital of Chongqing Medical University, Chongqing Key Laboratory of Ophthalmology and Chongqing Eye Institute, chongqing, China
  2. 2 University Eye Clinic Maastricht, Maastricht, Limburg, the Netherlands, Maastricht, Netherlands
  1. Correspondence to Peizeng Yang, The First Affiliated Hospital of Chongqing Medical University, Youyi Road 1, Chongqing 400016, P.R. China; peizengycmu{at}


Background Some uveitis subtypes show seasonal patterns. Whether these patterns are caused by seasonally varying temperatures or by other climatic factors remains unknown. This ecological research aimed to quantify the association between climate variability and uveitis onset.

Methods We combined data from the largest database of uveitis cases with surface climate data to construct panel data. We used choropleth maps to visually assess spatial uveitis variations.

Results Among 12 721 reports of uveitis originating from 31 provinces of mainland China from 2006 to 2017, we found that a 1°C increase in monthly temperature was associated with a rise in approximately 2 uveitis reports per 1000 individuals (95% CI 0.00059 to 0.0029). This association was present across all provinces, ranging in effect size from 0.0011 to 0.072 (95% CI 0.00037 to 0.10). A clear 0–3 months of cumulative lagging effect was noted across all types of uveitis, with the strongest effect for non-infectious uveitis (0.0067, 95% CI 0.0041 to 0.013). Stratified by age and sex, we found that men and people aged 20–50 years were more affected by temperature variations. Our model predicts that China might experience an increase in uveitis cases due to future global warming.

Conclusion Our study is the largest-ever investigation of the association between uveitis and climate and, for the first time, provides evidence that rising temperature can affect large-scale uveitis onset. These results may help promote and implement policies to mitigate future temperature increases and the burden of disease caused by global warming.

  • Epidemiology
  • Eye (Globe)
  • Public health

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Uveitis is considered as one of the leading causes of blindness in the world, with an estimated prevalence of 38–714 cases per 100 000 person-years and an incidence of 17–52 cases per 100 000 person-years.1 Although uveitis is generally thought to be an inflammation of the uveal tract (choroid, ciliary body and iris), it is broadly defined as inflammation not only of the uveal tract but also of related structures, including retina, sclera and orbit.2 More than 100 uveitis entities have been recognised.2 3 Genetic and environmental influences have been implicated in the development of uveitis. A robust genetic association has been observed with various uveitis entities, but the role of environmental influences on uveitis has not been thoroughly studied expect for cigarette smoking and vitamin D.2

Climate is increasingly being understood as an important environmental impact factor affecting many aspects of human health. Evidence from the past 20 years suggests that climate change, including global warming, can be associated with poor health outcomes.4 An increasing number of hypotheses potentially links climatic conditions and inflammatory diseases.5 6 It has been shown that some uveitis subtypes and related inflammatory factors have certain seasonal patterns,7 8 but no studies have determined whether the patterns are caused by seasonally varying temperatures, or by other seasonally varying climatic factors (such as precipitation, humidity and sunlight).7 9

It is important to determine whether uveitis is associated with climate, because it helps us understand this disease burden brought by changes in climate and may make a contribution towards appropriate political measures in mainland China, which is the world’s most populous country and largest emitter of greenhouse gases.10 11 Here, we quantified the associations between climate variability and the onset of uveitis by combining surface climate data (temperature as the main variable and other climate variables as control factors) and the data from the largest database of uveitis cases on a country-wide scale.



Data on uveitis were obtained from the uveitis centre of the First Affiliated Hospital of Chongqing Medical University. This uveitis referral centre has established a document management system which is currently the largest biological database of patients with uveitis in mainland China. The database reports the location, month, age, sex and diagnosis of all individuals with uveitis onset since May 2008.3 The management and diagnosis of these patients have been described in detail elsewhere.3 12–16 This study included patients with uveitis reporting to our centre with onset dates between January 2006 and December 2017. Cases with unknown uveitis onset month were excluded. Data extracted on the included cases (n=12 721) included the time (month) and location (county-level) of the onset, recording the monthly uveitis onset cases per county. These were matched with climate data of the same time from the nearest meteorological station, constituting together the climate–uveitis panel data. The data were analysed for all types of uveitis and then stratified by the following subsets as detailed in our previous report17: (i) aetiology (infectious vs non-infectious), (ii) type: idiopathic uveitis and two common uveitis entities (Vogt-Koyanagi-Harada disease (VKH) and Behçet’s disease (BD)) and (iii) uveitis alone or uveitis with a systemic disease. We further stratified the data by sex (male and female) and age (aged 20–50 years and others).1

Monthly surface climate data were retrieved from the National Meteorological Information Centre. These data are compiled from 613 ground meteorological observatories, and automated stations throughout China performed continuously since 1951. The reports are handled as follows: (1) the number of daily weather observations varies between counties. Following the unified regulations, daily average weather data are calculated based on measurements at 02:00, 08:00, 14:00 and 20:00. (2) Monthly average weather data are the average of all daily averages of that month, which includes temperature, relative humidity, atmospheric pressure, vapour pressure, sunshine hours and the monthly precipitation (online supplemental figure S1).

The unit root test and Engle-Granger test were used to analyse the data avoiding spurious regression. We assessed statistical significance at the 5% level.

Study area

Participants came from 613 counties. These counties are nested within the 31 provinces of mainland China: Beijing, Hebei, Shanxi, Tianjin, Inner Mongolia, Guangdong, Guangxi, Hainan, Anhui, Fujian, Jiangsu, Jiangxi, Shanghai, Shandong, Zhejiang, Gansu, Ningxia, Qinghai, Shaanxi, Xinjiang, Heilongjiang, Jilin, Liaoning, Chongqing, Sichuan, Guizhou, Tibet, Yunnan, Henan, Hubei and Hunan. Mainland China is located in the eastern Asia. Most of mainland China is in the temperate zone, while small parts in the tropical, subtropical and frigid zones. There are three levels of administrative divisions in mainland China: provincial, county and township.

Spatial variations in uveitis

The choropleth map was generated in the Geographic Information System (ArcGIS Version 10 software) to visually examine the geographical variations in the proportion of uveitis and geographic differences in the effects of temperature on uveitis in the provinces across mainland China. Quintiles were used to define map strata. The proportions included the following forms: (a) the provincial uveitis case numbers divided by the total number of cases (figure 1) and (b) the number of uveitis subtype cases in a province divided by the total number of cases in that province (online supplemental figure S2).

Figure 1

The proportion of patients with uveitis in mainland China by provinces, 2006–2017. We used quintiles to define map strata.


We followed the procedure in Climate Econometrics 18 and Panel Data Econometrics 19 to complete our models. Three models (fixed effect model, random effect model and pooling effect model) were constructed using the climate–uveitis panel data. The optimal model was selected finally based on the following three tests: the F test, the LM test and the Hausman test. We selected the fixed effect model (online supplemental table S1).

We estimated the following equation for our panel: Embedded Image 1

using ordinary least squares based on the chosen model (fixed effects regression model), where i indexes localities (counties), s indexes the province in which the locality is located, m indexes the month of the year and t indexes the year. We included Embedded Image and Embedded Image in equation (1) to ensure that unobserved characteristics of temporal and geographic factors did not disturb our assessment of the monthly temperature effect on uveitis. These terms are, respectively, time-invariant across-county differences and time-varying differences that account for unobserved characteristics constant across location and time.20 The disturbance term is Embedded Image , and Embedded Image represents monthly cases. The set of the climatic control variable terms is Embedded Image , with Embedded Image representing monthly average relative humidity, monthly precipitation, monthly average pressure, monthly average vapour pressure and monthly sunshine hours. Due to the exclusion of these control variables, our assessment of the effects of monthly temperature anomalies may be biased. The explanatory variable Embedded Image in equation (1) refers to the monthly average temperature in each county. Estimates in the fixed effects models can be equivalently interpreted as an absolute change in each unit of climate variables (eg, the effect of temperature that increases from 0°C to 1°C is equal to the effect from 30°C to 31°C). To estimate the potential lagged effects of temperature anomalies on future uveitis onset, we estimated a version of equation (1), where βL =1 and Embedded Image indicate the effect of the previous climate variables, and βL =0 and Embedded Image indicate the effect of the current climate variables. A finding of βL =1 and βL =0 suggests that high temperatures at a given month change the uveitis onset rate both during that month and in the following month. The sum of the coefficients βL=1 +βL=0 gives the overall effect of both months. We examined equation (1) across all-cause uveitis and uveitis subtype and then stratified our sample by age and sex.

To explore potential non-linear responses to temperature in our data, we estimated flexible versions of (T) , including higher-order polynomials: Embedded Image 2

The estimated model coefficients β provide the effect of the +1°C anomaly on the onset of uveitis.21

We further assessed whether the temperature–uveitis relationship was modified by the year or province, using equation (3): Embedded Image 3

where Embedded Image is a dummy variable for whether observation i falls into bin d. The coefficients provide the effect of +1°C separately for each bin (year dummies or province dummies).22

To calculate the potential effects of future climate change on uveitis, we used the Providing Regional Climates for Impact Studies (PRECIS) model to project climate change.23 The estimated greatest change in the cases of uveitis per +1°C deviation in temperature is Embedded Image , and ΔTut is the projected increase in temperature at the end of the 21st century (2099). The change in the cases of uveitis due to future temperature change is Embedded Image .


Characteristics of enrolled patients

The study population included 12 721 patients diagnosed with uveitis (table 1). Patients came from 31 provinces across mainland China with a relatively higher proportion coming from Chongqing and Sichuan (figure 1). The male-to-female ratio was 1.11. The most common onset age of uveitis was between 20 and 50 years1 and was diagnosed in 8229 patients (64.7%). There were 46.3% individuals diagnosed with idiopathic uveitis, and the remaining were classified with other specific entities, among which VKH accounted for 13.1% and BD accounted for 9.9%. BD and VKH are the two most frequently diagnosed clinical uveitis entities in mainland China.24–26 Of the cases, 67.7% were patients with uveitis where the disease was limited to the eye, and 32.3% were patients with uveitis having a systemic disease. Infectious uveitis accounted for 3.7% and non-infectious uveitis accounted for 96.3% of cases, respectively (table 1).

Table 1

Characteristics of enrolled patients

Spatial variations

Figure 1 demonstrates the spatial variations in the proportion of patients with all types of uveitis at the provincial level, stratified into quintiles (figure 1 and online supplemental table S1 and figure S2). Chongqing and Sichuan were in the highest quintile for the proportion of uveitis cases, whereas Tibet, Hainan, Tianjin, Beijing, Shanghai, Qinghai, Guangxi, Inner Mongolia and Hebei were in the lowest proportion quintile. The proportions ranged from 0.23% to 26.09%.

Historical effect of rising temperatures on uveitis

The results indicated that a 1°C increase in monthly temperature produces an increase of about two uveitis cases per 1000 individuals (0.0017, 95% CI 0.00059 to 0.0029; figure 2 and figure 3). In sensitivity analysis, the quadratic effect of temperature on uveitis was not significant and its inclusion did not alter the linear effect of temperature on uveitis to an important degree (online supplemental figures S2 and S3 and table S2).

Figure 2

Effects of temperature on the number of uveitis cases. The lines show the estimated relationship between monthly temperature and monthly uveitis cases. The blue shaded areas are the bootstrapped 95% CI on model 1 (online supplemental table S2). The histograms at the bottom display the distribution of monthly temperatures across sample locations.

Figure 3

Effect of variation in temperature on number of uveitis cases and temporal displacement across the full sample and sub-groups. (A, B) The dots are point estimates of the effect of monthly temperature on monthly uveitis cases (from equations (1), (3) and (4)); the lines are 95% CIs. lag 0, current month; lag 1, 1-month lag; lag 2, 2-month lag; lag 3, 3-month lag.

Heterogeneous effects and adaptation

There were positively significant associations between temperature and uveitis for each province (figure 4A). The strength of temperature–uveitis association ranged from 0.0011 to 0.072 (95% CI ranged 0.00037 to 0.10) in provinces of mainland China. Furthermore, we found that the association between temperature and uveitis became stronger in the course of our study period (figures 3A and 4B). In addition to the effects of provinces and years, we may expect that not all individuals would be similarly affected by anomalous increases in temperatures. Compared with all patients, we found similar but slightly weaker effects of temperature in patients aged 20–50 years (0.0013, 95% CI 0.00050 to 0.0021; figure 3B). Next, there was a significant difference in the effects of temperature on uveitis by sex: temperature anomalies have a significant effect on the onset of male uveitis (0.0014, 95% CI 0.00066 to 0.0021), but not on female uveitis (figure 3B). Moreover, there was a statistical significance in the effects of hotter temperatures on uveitis with systemic diseases (0.00092, 95% CI 0.00041 to 0.0014) and non-infectious uveitis (0.0016, 95% CI 0.00053 to 0.0028) (figure 3B).

Figure 4

Temperature effects on uveitis over time and space. (a) Temperature effects over space. The colours show the increases in the province-specific monthly uveitis cases per 1 °C increase in monthly temperature. (b) Temperature over time. Each dot is the year-specific effect of temperature on uveitis (the line is the 95% CIs). The red dashed line shows the average effect across the full sample.

Lagged effect of rising temperatures on uveitis

We found an association between abnormally hot temperatures in the previous months and uveitis occurrence in the current month, with a lag of up to 3 months (figure 3). The greatest effect was at a lag of 2 months (0.0027, 95% CI 0.0016 to 0.0039). The strength of the cumulative effect of temperature over lags of 0– 3 months on all types of uveitis was 0.0093 (95% CI 0.0046 to 0.014). Stratified by uveitis subtype, the cumulative effect was strongest for non-infectious uveitis (0.0067, 95% CI 0.0041 to 0.013), followed by uveitis alone (0.0050, 95% CI 0.0022 to 0.0079), uveitis with systemic diseases (0.0034, 95% CI 0.0014 to 0.0055), idiopathic uveitis (0.0030, 95% CI 0.00092 to 0.0050), VKH (0.00081, 95% CI 0.00018 to 0.0014) and infectious uveitis (0.00043, 95% CI 0.000089 to 0.00076). There was no lagged effect of temperature on BD. Furthermore, temperature anomalies had a cumulative effect over lags of 0–3 months on male individuals (0.0050, 95% CI 0.0022 to 0.0079), female individuals (0.0038, 95% CI 0.0016 to 0.00060), individuals aged 20–50 years (0.0062, 95% CI 0.0029 to 0.0095) and other ages (0.0027, 95% CI 0.00088 to 0.0045) (figure 3B).

Potential implications of climate change for uveitis

Our historical data revealed that abnormally high past temperatures significantly altered the incidence of uveitis. In addition, we found that the annual average temperature in 2017 was 0.17°C higher than in 2006, which is consistent with global warming. To calculate the potential impacts of future climate change on uveitis, climate projections using the PRECIS model suggested that the annual mean temperature in mainland China will increase by up to 6°C at the end of 21st century.23 The standard assumption for climate impact is that outcomes in the future will respond to hot and cold months in the same manner as past outcomes, and applying future changes in annual mean temperature (ΔTut) to the temperatures–uveitis coefficient (Embedded Image ) is appropriate.21 27–29 Based on this assumption and absent adaptation, we calculated an increase of approximately 16 uveitis cases per 1000 individuals in mainland China by 2099.


We provide longitudinal country-scale evidence that uveitis onset in mainland China is tightly associated with the local temperatures. This association gets stronger over time and is highly significant across all 31 provinces. As expected, we found clear evidence for a cumulative lagged effect over 0–3 months in all types of uveitis, with the strongest effect on non-infectious uveitis. Stratified by age and sex, our results indicate that risks from abnormal temperature exposure are higher for male individuals and people aged 20–50 years. We further predict that China might experience an increase in the uveitis onset rate due to future climate change-induced warming.

These findings are consistent with previous studies on close links between human health and climate.5 Although a growing body of literature has revealed the effects of temperature on morbidity and mortality,30–33 to our knowledge, ours is the first study to examine the effects of temperature on uveitis. We jointly analysed all cases of uveitis and provided a comprehensive overview of the temperature–uveitis association. Our results do suggest that the association is getting stronger over time. However, these results contrast with common reports that the relationship between temperature and health has decreased over time.18 34–36 This discrepancy may be because mainland China has become more industrialised and produces larger greenhouse effects or that other aspects of development increase environmental exposure over time.

Although our research and previous reports have suggested that the geographical background of uveitis is different,1 37 the impact of temperatures on uveitis is similar for most of the provinces in this study. It does support that effects in provinces with hotter average temperatures are indistinguishable from effects in cooler regions. This means that individuals who are more frequently exposed to hot temperatures are not less sensitive. This conclusion starkly contrasts with the dominant themes in the existing climate-health literature. These previous reports suggest that the effects on health increase in hot temperatures and decrease in cold temperatures.18

We do find clear evidence that the effects of temperature are only evident among those who are most commonly affected by uveitis in developing countries.1 Our results indicate that risks from abnormal temperature exposure more often affect male patients and patients aged 20–50 years, but the effects are smaller than on all patients in our study. Our results suggest that the effects of temperature on all uveitis cases are higher than on specific uveitis subtypes (uveitis with systemic diseases and non-infectious uveitis). Although there are obvious differences in the effects of temperature by sex, age and type in the current month, we find that the temperature effects lagged in almost all subgroups. Our results suggest that abnormally hot temperatures can cause additional uveitis cases across the full sample and in different subgroups except BD. There may be other factors that play a more prominent role in the pathogenesis of BD rather than ambient temperatures. Previous laboratory studies have revealed that hereditary factors and certain environmental factors (vitamin D metabolism) are thought to be involved in the development of BD.2 In addition, the cumulative effect over lags of 0–3 months was strongest on non-infectious uveitis. A biologically plausible mechanism is that climatic variables might primarily influence the autoimmune response, inducing continuous production and accumulation of autoimmune-related inflammatory factors, eventually leading to immune-related uveitis, such as non-infectious uveitis. Interestingly, contradictory to our results, a recent report38 that has addressed the weather influence on uveitis in a much smaller area in Southern Europe found uveitis to occur more frequently under rainy and windy conditions. The reason for this difference might be that the uveitis aetiology in their sample was very different from ours. They found B27-related and herpetic uveitis to be the most common aetiologies, with a higher incidence in women rather than men. This might also represent a different response to meteorological factors across different diseases or geographical areas.

Nevertheless, these uveitis events cannot be found in the effects of the abnormally hot temperatures in a future month on the current month. We further calculated projected changes in the onset of uveitis resulting from future climate change. Global warming due to climate change can increase the onset of uveitis in the future. Future climate change projections are predicted relative to 2005. Although it is a little different from our years (post-2006), the effect size itself is an estimate of the approximate value. Positive feedback loops from other potential effects, such as environmental pollution and acute trauma, may amplify the relationship.

Our study has several limitations. First, this study suggests that temperature is associated with the onset of uveitis, but the specific mechanisms behind this association remain unclear. For example, temperatures might affect the production and secretion of pro-inflammatory factors in the body. Alternatively, indirect effects of climate change or variability, such as psychological stress, might be involved since the local population is not used to high temperatures and does not know how to cope with it. Second, the study was done on patients treated at a referral centre; therefore, there is a selection bias in the patient population investigated. Third, as an ecological retrospective study, ecological fallacies and temporal bias might be involved. Fourth, although the samples were divided according to the uveitis subtype, some subtype cases were insufficiently described and thus could not be further analysed. Finally, the applicability of the results is not universal. Since the climate data came from China, the conclusion might be limited to countries with similar infrastructure and climate. Further analyses are necessary to determine if these results are applicable to the rest of the world.

In summary, our study is the largest investigation to date on the relationship between uveitis and ambient temperature and, for the first time, provides evidence that temperature increases can significantly affect the onset of uveitis. Our study will help promote and implement policies to mitigate future temperature increases and the burden of disease caused by global warming.



  • HT and SP contributed equally to this work.

  • Contributors PY and HT conceived and designed the study. HT, SP, ZZ, JS and WL collected clinical data. HT and SP analysed and interpreted the data. HT wrote the first draft of the paper. PY, GS and KA reviewed and edited the manuscript. All authors provided a final review and approved the manuscript before submission.

  • Funding This study was supported by Chongqing Outstanding Scientists Project (2019), Chongqing Key Laboratory of Ophthalmology (CSTC, 2008CA5003), Chongqing Science & Technology Platform and Base Construction Program (cstc2014pt-sy10002) and the Chongqing Chief Medical Scientist Project (2018).

  • Map disclaimer The depiction of boundaries on the map(s) in this article does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. The map(s) are provided without any warranty of any kind, either express or implied.

  • Competing interests None declared.

  • Ethics approval The experimental procedures and research design were conducted in accordance with the tenets of the Declaration of Helsinki. The Clinical Research Ethics Committee of the First Affiliated Hospital of Chongqing Medical University approved this study and waived the need for informed consent based on the retrospective nature of the study and anonymisation of the data.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement All data relevant to the study are included in the article or uploaded as supplemental information.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Linked Articles

  • At a glance
    Frank Larkin