The neglected tropical disease (NTD) trachoma, caused by Chlamydia trachomatis, is the world's leading cause of infectious blindness [1] and an important cause of chronic discomfort in 57 endemic countries, mainly in Africa [2]. Over 40 million people are infected with C. trachomatis, 8 million of whom have painful, in-turned eyelashes (trichiasis) [3]. It is estimated that 1.2 billion people live in trachoma-endemic areas [4]. The current distribution of trachoma aligns with low and middle-income countries, within which poorer individuals and communities are at highest risk [2], [5].

The World Health Organization (WHO) classifies trachoma into active and chronic stages. The signs of active trachoma, according to the simplified grading system [6], are trachomatous follicles (TF) and trachomatous inflammation (TI). The chronic, potentially blinding stages of trachoma are characterised by visible scarring of the under surface of the upper eyelid (TS), in-turned eyelashes, trachomatous trichiasis (TT), and corneal opacity (CO). The incidence of chronic trachoma increases with age. Active infection principally occurs in children, where it is self limiting, whereas the blinding stages of trachoma are not seen until later in life. Although repeated episodes of infection during childhood are thought to lead to the scarring stages of trachoma, the natural history and pathophysiology are not fully understood. Both clinical presentation and transmission of trachoma are complex and multi-faceted. The investigation and monitoring of active trachoma are complicated as evidence of infection with Chlamydia does not correlate well with clinical signs of the disease (i.e. TF and TI) [7].

Trachoma is transmitted via contact with infected eye and nasal secretions by hands, fomites, and eye-seeking flies [8]. Individual factors such as rural residence, overcrowded living conditions, and keeping cattle close to the home are also associated with trachoma [5], [9]–[11]. The strongest environmental risk relates to poor hygiene, often reflecting poor access to water and lack of sanitation, which promotes transmission by providing more effective breeding sites for eye-seeking flies [12]–[17]. The strategy for trachoma control is the SAFE strategy: S = surgery to correct the upper eyelid deformity [18]; A = antibiotics for active infection via mass drug administration (MDA) [19]; F = facial cleanliness [20], and E = environmental improvement [21]. The ‘F’ and ‘E’ elements aim to reduce C. trachomatis transmission by improving hygiene behaviour and reducing environmental factors which promote eye-seeking flies.

There has been growing interest in the direct effects of climate on disease, reflecting concerns about climate change as well as improvements in the use of climate information for disease control [22]. Epidemiological studies of weather (daily temperature or rainfall) use time series methods to detect acute (short-term) effects on health outcomes. Long term exposures (i.e. climate) have also been studied in cross-sectional studies [23]. Climate factors vary in time and space and it is important that the study design includes appropriate adjustment for social or environmental factors (confounders). Associations between meteorological variables and health outcomes are likely to depend on local contexts, and have also been shown to change over time [22]. Climate is the average weather conditions observed over a long time period (decades). Globally, the world can be divided into five main climate types (with multiple sub-types) based on annual average and monthly temperature and precipitation values. The best known classification scheme is the Koppen-Geigen system which describes the following climate types: tropical; dry (arid and semiarid); mild temperate; continental and polar [24].

Climatic factors may influence the distribution and prevalence of trachoma indirectly through poor access to water, which limits hygiene behaviour, or low rainfall, which may influence the distribution, abundance or seasonal activity of Musca sorbens, the principal eye-seeking fly implicated in trachoma transmission. At broad geographic levels, climate also influences agricultural productivity and livelihoods in resource-poor settings; communities at risk of trachoma often depend on livestock and subsistence farming. Direct influences on the eye may include humidity, dust and aridity.

The purpose of this study was to examine the association between climatic factors and the distribution, frequency and severity of trachoma. A systematic literature review was conducted for evidence of climatic effects on active trachoma, to identify factors associated with transmission, as well as cicatricial/blinding trachoma, to identify factors associated with scarring.

Searches were undertaken separately for active and chronic trachoma (hereafter termed trachoma outcomes) (See search terms in Supplementary Material Table S1). The following electronic databases were searched: CAB Abstracts; Embase; Global Health; Medline; Web of Science. Websites of international agencies were also searched: the World Health Organization (WHO), WHO Special Programme for Research and Training in Tropical Diseases (TDR); Intergovernmental Panel on Climate Change (IPCC); United Nations Children's Fund (UNICEF); UN-Habitat; The Carter Center; the International Trachoma Initiative (ITI); Sightsavers; Helen Keller International; Fred Hollows Foundation; Christian Blind Mission.

Studies were only included if they quantified an association between a climate factor (temperature, rainfall, altitude, etc) and a trachoma outcome. Papers were not excluded based on geographic location of study, age of participants, or language of journal publication. Peer reviewed journal articles and reports from leading international agencies (e.g., WHO) published between 1 January 1950 and 1 April 2012 were included.

Data extraction and analysis were conducted by two readers. Results were screened and reviewed in three stages: i) title; ii) title and abstract; and iii) title, abstract, and manuscript. References for which title and abstract were available and which seemed to meet the inclusion criteria were reviewed by two reviewers for eligibility, quality and data extraction. The quality of each study included was assessed using the parameters described in the STROBE checklist for cross-sectional studies [25]. The quality of observational studies was assessed using the following criteria: sampling of study population to estimate trachoma prevalence, study design and control of confounding, reporting of negative results, and the measurement of climate exposures.

It was expected that heterogeneity in the study designs and exposure measurements would preclude meta-analyses. We also reviewed the evidence for climate type and seasonality on trachoma outcomes.

Only eight of the 1751 papers retrieved met the inclusion criteria (Figure 1, Table 1) [17], [26]–[32]. All studies were cross-sectional and all used the WHO simplified grading system for trachoma, which requires use of a ×2.5 magnifying loupe while examining the under surface of the upper eyelid to give more detail of pathological changes. However, use of magnification was inconsistent or not reported. All eight studies had been published since the year 2000, although field work in one study had been undertaken earlier [28]. All were undertaken in African countries, principally across the Sahel belt. Three of the eight studies were conducted in West Africa: Mali, two [26], [28]; Burkina Faso, one [29], and five in East Africa: Ethiopia, three [30]–[32]; Tanzania, one [17] and South Sudan, one [27].

Four of the eight studies were judged to be of moderate quality [26]–[29] and four of lower quality [17], [30]–[32]. The four moderate quality studies provided the best and most comprehensive measurement of climatic exposures, used a more detailed methodology and complex analyses. Three of these studies used epidemiologically robust population based methods to assess the prevalence of trachoma [26]–[28], two of which limited analyses to acute trachoma in children [27], [28]. The fourth study used a school based sampling strategy of children aged 7–11 years [29]. The moderate quality papers all measured several climatic factors using data from weather stations (two studies) or gridded datasets (two studies). Moderate quality papers also tended to account for the possible function of time, allowing for trend and seasonal effects, and reported negative results. The lower quality papers all used robust methods to assess the prevalence of trachoma, but only assessed altitude and one did not take sufficient account of confounding [32]. All eight papers reported results for active trachoma, but only one reported a finding for chronic trachoma in adult women in Mali [26].

Two further studies that provided limited evidence of the effects of climate zone on trachoma are summarised in Table 2 [11], [33] and three further papers reported on seasonality [34]–[36].

Studies designed to explore associations between climate and disease should ideally use data on temperature and precipitation that is valid for the population under study, for example, from weather station observations. For large populations and areas, gridded data from weather stations or satellite data might have to be used [38]. The latter are particularly relevant in sub-Saharan Africa where coverage by weather stations is low, although care should be taken in highland areas, where use of wrong climate data can give spurious results [39]. Altitude is often used as a crude proxy for temperature as temperatures decrease with increasing altitude. However, altitude also influences a number of other factors, including rainfall; highland areas may be wetter or drier than surrounding low land areas depending on the local climate and topography. Furthermore, vegetation, land use, and population characteristics can also vary by altitude. It is therefore important that studies assessing the influence of altitude on disease take account of these potential confounders. Temperature is also a more robust indicator than rainfall over a geographic area, due to the higher spatial and temporal variability in precipitation. Meteorological variables are often correlated in space and time and therefore it is important that the most appropriate parameterizations (including meteorological indices) are decided a priori.

This systematic literature review found evidence that low rainfall is associated with a higher prevalence of active trachoma, which is consistent with the finding that trachoma prevalence is greater in savannah areas. In one study, however, rainfall was removed from the final multivariate model due to collinearity with latitude and humidity. Low rainfall warrants further investigation as a risk factor for the distribution and prevalence of trachoma.

Despite a wealth of anecdotal information, there is very little high quality observational evidence on the role of temperature on the distribution or prevalence of active trachoma. There is even less evidence regarding chronic trachoma (and related blindness). One study reported a significant effect of lower latitude on trachomatous scarring and trichiasis in adult women in Mali [26]. Using altitude as a proxy for temperature, there is some evidence that in highly endemic areas of East Africa the prevalence of trachoma is lower at high altitudes but two papers, both risk mapping papers which also measured several co-linear variables (e.g., rainfall, humidity) did not find statistically significant associations. Although it is plausible that low temperatures have a limiting effect on the distribution of trachoma, the association could be due to confounding by social or other environmental factors. For example, in Kenya it has been observed that households of higher socio-economic status and less over-crowding tend to reside at higher altitudes than poorer households [17].

The activity and distribution of eye seeking flies may be a mechanism by which low temperatures limit trachoma prevalence, as climatic factors directly affect the seasonal activity and distribution of Muscid flies (e.g., M. sorbens). Laboratory studies have also shown that the lifespan of M. sorbens ranges from less than 12 days at 32°C to 35 days at 24°C [29], [40]. Field studies show that the distribution of M. sorbens is strongly correlated with altitude [41]. Climate may also effect fly activity indirectly, for example, through access to faeces for breeding, as high temperature and sunshine may induce rapid drying of faecal matter, rendering it a less effective breeding site [29].

Historically, trachoma studies have focused on changes in distribution (e.g., amongst migrant populations) and few doubted the association between dust, heat, and trachoma [42]–[44]. The systematic review found no studies that quantified an association between dust and trachoma, probably because it is difficult to measure dust exposures, especially across large spatial scales.

Only one paper reported a seasonal pattern in active trachoma [34]. Assessing seasonality in relation to trachoma is problematic for several reasons. Not only is there poor correlation between PCR-confirmed Chlamydial infection and the clinical signs of active trachoma, but the interval between infection and the development and resolution of clinical signs is not yet known. It is therefore important to evaluate the method of assessment of active trachoma when interpreting temporal patterns of the disease.

This review has several limitations. The quality of the climate exposure assessment in many studies was poor due to the use of broad categories (e.g. altitude) or lack of local measures for temperature and rainfall, due to limited coverage by weather stations. However, the Sudanese study used satellite data to obtain comprehensive exposures of average rainfall and temperature at a reasonable resolution, although the data would have been interpolated leading to some parameter uncertainty [27]. All the studies were undertaken in Africa, which limits generalisability of the findings. It is possible that climate-trachoma associations are location specific. For example, the relationship between trachoma and altitude may be different in Asia where overcrowding during the winter among populations living at high altitude is thought to promote transmission (B. Qureshi, personal communication). Other limitations are that some studies used relatively small sample sizes and prevalence estimates of trachoma had wide 95% confidence intervals, which reduce the power to detect statistically significant differences in outcomes. Most trachoma research is conducted either within foci where trachoma is endemic, or is undertaken to identify trachoma endemic areas where control programmes are required. This is a major limitation when exploring climatic factors where the edges of the distribution are of interest. It is also possible that some publication bias has occurred, with studies not finding an association not being published.

The current understanding of trachoma transmission is that multiple factors combine to propagate this preventable disease. Many of these factors are difficult to quantify, but are very important when considering the entire transmission cycle and interactions between socio-economic and environmental factors. The findings of this study suggest that climatic factors may also play a role in the distribution and prevalence of trachoma. Socioeconomic factors (e.g., poverty) and certain behaviours (e.g., migration) warrant further attention as they impact on those at risk of trachoma and are, in turn, affected by environmental factors [5], [45], [46]. The findings of this review support the call for greater investment in the “E” element of the SAFE strategy.

The WHO-led Alliance for the Global Elimination of Blinding Trachoma (GET2020) aims to eliminate blinding trachoma by the year 2020, and there are international partnerships, including a drug donation programme for control [4], [47], [48]. There are ongoing international efforts to map several NTDs, including trachoma, providing up to date information on the distribution of trachoma. The findings of this review are of value for those mapping the distribution of trachoma, as altitude, temperature and rainfall may be additional parameters for consideration at the planning stages. Better delineation of trachoma endemicity, which will identify the edge of endemic foci, will allow more informative studies of the impact of climate over a large scale on the distribution of trachoma.

The findings of this review add impetus to trachoma control because if the climate in sub-Saharan Africa was to become hotter and drier (due to either natural variability or anthropogenic forcing), this may potentially influence the distribution and severity of trachoma. Finally, this review highlights the relative paucity of studies exploring these potential associations and the poor quality of much of the climate data both in terms of coverage and frequency at which the data were collected.