Do temperature changes cause eczema flares? An English cohort study

Abstract

Background

It is unclear if ambient temperature changes affect eczema. It is also unclear if people with worse disease are more susceptible to weather-related flares, or specific types of emollient offer protection.

Objectives

To investigate the effect of short-term temperature variations on eczema symptoms in children.

Methods

Data from a UK cohort of 519 children with eczema were combined with data from the Hadley Centre’s Integrated Surface Database. Hot and cold weeks were defined by average regional temperature > 75th or < 25th percentile, January 2018 to February 2020. Eczema flares were defined as ≥ 3-point change in Patient-Oriented Eczema Measure (POEM). Random-effects logistic regression models were used to estimate the odds ratios of flares in hot and cold weeks (reference group: temperate weeks).

Results

The baseline mean age was 4.9 years (SD 3.2) and the POEM score was 9.2 (SD 5.5). From the 519 participants, there were 6796 consecutively paired POEMs and 1082 flares. Seasonal variation in POEM scores was observed, suggesting symptoms worsening in winter and improving in summer. Odds ratios of flares were: 1.15 [95% confidence interval (CI) 0.96–1.39, P = 0.14] in cold weeks and 0.85 (95% CI 0.72–1.00, P = 0.05) in hot weeks. The likelihood ratio test showed no evidence of this differing by disease severity (P = 0.53) or emollient type used (P = 0.55).

Conclusions

Our findings are consistent with previous studies demonstrating either improvements in eczema symptoms or reduced flares in hot weather. Worse disease and different emollient types did not increase susceptibility or provide protection against temperature changes. Further work should investigate the role of sunlight, humidity, pollution and other environmental factors.

What is already known about this topic?

• People with eczema commonly report that changes in the weather cause disease flares.

• Previous studies have suggested that some patients (such as those with more severe disease) are more susceptible to changes in the weather.

• Some guidelines recommend thicker emollients in winter months.

What does this study add?

• Hot, but not cold, weather was associated with reduced odds of a flare in eczema in children.

• Susceptibility to changes in temperature was not associated with eczema severity or the type of emollient used by the child.

• Specific types of emollients may not protect against changes in temperature.

Eczema (atopic eczema/dermatitis) is an inflammatory dry itchy skin condition affecting around 20% of children, commonly persisting into adulthood.¹ Eczema waxes and wanes in severity, with ‘flares’ caused by different ‘triggers’. Identifying triggers can be difficult as there is variation among patients, but some of those previously investigated include the climate, weather, diet and clothing.^2,3 Clarifying triggers is important to patients as it can help inform self-management strategies and reduce the need for rescue treatments.

Weather describes the conditions of the atmosphere over short periods of time, ranging from minute-to-minute changes to periods lasting weeks. This differs from the climate, which describes weather patterns on average over the long term, usually years to centuries. Previous epidemiological studies have primarily focused on associations between eczema severity/prevalence and climate.^4–7 Some smaller studies in countries with temperate weather similar to the UK have provided varied findings around short-term effects. An Irish cohort study of 25 children with eczema demonstrated worse scratch scores associated with high outdoor temperature.⁸ A follow-on study of 60 children in Nottingham (England) showed an increase in eczema severity with shampoo use (particularly when the outdoor temperature was low).¹ No other associations were seen between eczema and temperature or humidity. A nationwide questionnaire sent to 1343 Danish outpatient patients with eczema between 2014 and 2018 reported associations by eczema severity.⁹ Of those with mild eczema, 47% reported cold weather and 19% warm weather as causes of worse eczema. In children with severe eczema 62% reported cold weather and 45% warm weather as triggers. A larger proportion of children with severe eczema reported weather as triggers, suggesting they are more prone to weather-related flares. This is possibly because of a greater defect in the functioning of the skin barrier compared with those with milder disease.

Overall, there is limited information about the effects of short-term weather on eczema flares in the literature. Exploring this with larger cohorts than previous studies and quantitative methods is important to further our understanding of the potential temperature triggers of eczema, as well as potential mitigatory effects of different emollients. This exploratory study serves as a step towards providing new scientific information about weather-related eczema in the UK.

Patients and methods

We sought to investigate the effect of weather (temperature)-related changes on eczema flares in children living in England, which has a temperate climate. Our objectives were to (i) describe the seasonal trend in eczema symptoms; (ii) explore associations between changes in temperature and eczema severity; and (iii) explore for associations between baseline eczema severity and emollient type used with temperature-related eczema flares.

We used data from a randomized controlled trial of emollients for the treatment of eczema in children, the Best Emollient for Eczema (BEE) study.¹⁰ In BEE, 550 children with eczema were randomized to use a lotion, cream, gel or ointment emollient as their main moisturizer for 16 weeks. No difference was found in the primary outcome of POEM (Patient-Oriented Eczema Measure), which captures eczema symptoms for the previous week.¹¹

Children were recruited from general practitioner (GP) surgeries in three regions of England: West of England, East Midlands and Wessex. Eligible participants were aged 6 months to 12 years of age, with a diagnosis of eczema on their GP record, of at least mild severity (POEM > 2) and their parent was able to complete the outcome measures. Patients were excluded if they had any allergies to the study emollients. Recruitment began in January 2018 and weekly data (for 16 weeks) collection ended in February 2020.

Temperature data

Daily mean dry bulb temperatures were obtained from HadISD (Met Office Hadley Centre Integrated Surface Database).¹² Dry bulb temperature is defined as ambient air temperature measured using a thermometer shielded from moisture.¹³ HadISD weather data are provided daily at the city level, collected from local weather stations and quality controlled by the UK Met Office to be consistent across the whole of the UK. Using the Stata geodist command (StataCorp, College Station, TX, USA), participants were mapped by their home post codes to their nearest weather station. Seven-day rolling averages for temperature were calculated and linked to participants’ corresponding POEM score for that week. We also classified 7-day moving averages as either hot, cold or temperate weeks. Hot weeks were defined as those with average temperature greater than the 75th percentile, temperate weeks between the 25th and 75th percentile and cold weeks less than the 25th percentile of the temperature distribution during the total observation period (see Table S1 in the Supporting Information). These cutoff points were used to ensure a reasonable number of POEMs in each category. This technique has been previously used by other epidemiological studies in this field when comparing continuous variables with a dichotomous outcome.⁷

Definitions of eczema severity and disease flares

Eczema severity categories used in our descriptive analyses are those that Charman et al. have previously described using POEM score cutoffs. These are 0–2 (clear/almost clear); 3–7 (mild); 8–16 (moderate); 17–24 (severe); and 25–28 (very severe).¹¹ There is no consensus about how best to define an eczema flare.¹⁴ However, studies of POEM have identified a change of 3 as a minimally important change (MIC).^15,16 Therefore, we defined an eczema flare as an increase in POEM score ≥ 3 from the previous week.

Sample size

The BEE trial was powered to detect a difference of three units in POEM scores between treatment arms; full details of the assumptions for the sample size calculation can be found in the published BEE protocol.¹⁷ As an exploratory secondary analysis of a clinical trial this study did not set out to formally test a hypothesis but to explore relationships between variables and so no formal sample size was calculated.

Statistical methods

Descriptive analyses were undertaken of baseline characteristics such as age and eczema severity. This summarized numeric data using means and standard deviations. Categorical data used frequencies and proportions. The weekly measures of flares were studied using univariable and multivariable random-effects logistic regression models to estimate the odds ratios (ORs) of flares in hot and cold weeks using temperate weeks as the reference group. Multivariable analyses adjusted for age at baseline, emollient allocation, baseline severity (POEM at enrolment), gender, ethnicity and socioeconomic status. Sensitivity analyses were performed with varying cutoffs defining hot and cold weeks (see Table S1 in the Supporting Information). To explore whether the effect of weather differed according to emollient allocation or by baseline eczema severity, the primary multivariable regression models were run including an interaction term with weather. The likelihood ratio test was then used to compare this model with one excluding the interaction term to assess for evidence of effect modification. All analysis was performed using the statistical software Stata v17.0.

Results

Baseline characteristics

From the 550 participants in the original BEE trial dataset, we restricted our analysis to the cohort of 519 participants who had at least 1 pair of consecutive POEM measurements, to allow calculation of ‘flares’. These 519 children comprised 53.0% boys, 87.0% of White ethnicity with a mean age at enrolment of 4.9 years. Our cohort was similar to participants in the trial (Table 1) and a range of eczema severities were represented; clear or mild eczema (42.0%), moderate eczema (46.8%) and severe or very severe (11.0%). The mean baseline POEM score was 9.2 (SD 5.5). The majority (89.8%) of participants were within 20 km of their local weather station.

Table 1

Participant characteristics at baseline

Characteristic	BEE trial	Study cohort
Participants	550	519
Age, years; mean (SD)	4.9 (3.20)	4.9 (3.24)
Boys	295 (53.6)	275 (53.0)
Girls	255 (46.4)	244 (47.0)
Baseline eczema severitya
Clear/almost clear eczema (POEM 0–2)	40 (7.3)	40 (7.7)
Mild eczema (POEM 3–7)	185 (33.6)	178 (34.3)
Moderate eczema (POEM 8–16)	266 (48.4)	243 (46.8)
Severe eczema (POEM 17–24)	53 (9.6)	52 (10.0)
Very severe eczema (POEM 25–28)	5 (0.9)	5 (1.0)
Missing	1 (0.2)	1 (0.2)
Ethnicity
White	473 (86.0)	450 (86.7)
Black	18 (3.3)	15 (2.9)
Asian	16 (2.9)	15 (2.9)
Mixed	43 (7.8)	39 (7.5)
Socioeconomic background (IMD quintiles)b
IMD 1	62 (11.3)	57 (11.0)
IMD 2	55 (10.0)	48 (9.2)
IMD 3	102 (18.5)	96 (18.5)
IMD 4	111 (20.2)	105 (20.2)
IMD 5	173 (31.5)	169 (32.6)
Missing	47 (8.5)	44 (8.5)
Within 20 km of weather station	491 (89.3)	466 (89.8)
Distance to nearest weather station, km; mean (SD)	11.17 (7.4)	11.02 (7.3)

Data are n (%) unless otherwise indicated. POEM, Patient-Oriented Eczema Measure. ^aCategorized POEM score; ^bIndex of Multiple Deprivation (IMD), 1 is most deprived.

Weather

During the 25-month study period, the mean rolling 7-day average temperature was between 10.0 °C and 11.4 °C in the three regions (see Table S1 in the Supporting Information). The coldest week saw an average temperature of –2.4 °C in the south west, whereas the hottest week had an average high of 22.6 °C in the south coast area. The year 2018 had a notably cold winter because of anomalous atmospheric conditions, colloquially known as the ‘Beast from the East’. Temperatures were broadly similar across all three regions with similar peaks in summer and dips in winter (Figure 1).

Figure 1

Seven-day rolling average temperatures across three regions of England over the study period. The horizontal dotted lines indicate the 25th and 75th percentiles of the respective temperature distributions. For the combined figure, south west percentiles are used for the dotted lines for demonstrative purposes. HadISD, Hadley Centre Integrated Surface Database.

Patient-Oriented Eczema Measure scores

In total 87.6% (7727/8823) of weekly POEMs were completed, with 6796 pairs of consecutively completed POEMs. The number of completed POEMs varied across the study period, reflecting the number of participants in the trial. Some months therefore had more data available than others (Figure S1, see Supporting Information). Plotting of the average weekly POEM score of the overall cohort by month suggests a seasonal variation in eczema; worse in winter, better in summer (Figure 2).

Figure 2

Plot of average cohort Patient-Oriented Eczema Measure (POEM) score and south west temperatures by month.

Primary outcome: odds of eczema flare

In total, 1082 flare episodes were identified with 79.6% (413/519) of children having at least one flare. When comparing hot weeks with temperate weeks we found that the former were associated with a reduced odds of a flare [unadjusted OR = 0.83, 95% confidence interval (CI) 0.71–0.97, P = 0.02, Table 2]. This association was largely unchanged after adjusting for age, gender, ethnicity, socioeconomic status and baseline POEM in multivariate regression (adjusted OR = 0.85, 95% CI 0.72–1.00, P = 0.05). Cold weeks, conversely, were associated with increased odds of a flare (adjusted OR = 1.15, 95% CI 0.96–1.39, P = 0.14); however, the confidence interval overlapped with the null.

Table 2

Random-effects logistic regression: odds ratio of flare in hot and cold weeks, using temperate weeks as reference group^a

Week	Contributing (consecutive) POEMs, n	Unadjusted			Adjustedb
		Odds ratio	95% CI	P-value	Odds ratio	95% CI	P-value
Cold	1192	1.13	0.94–1.35	0.19	1.15	0.96–1.39	0.14
Temperate	3315	1.00	–	–	1.00	–	–
Hot	2289	0.83	0.71–0.97	0.02	0.85	0.72–1.00	0.05

CI, confidence interval; POEM, Patient-Oriented Eczema Measure. ^aHot and cold weeks defined by 25th and 75th percentiles for each region (Table S1; see Supporting Information). ^bAdjusted for age at baseline, POEM at baseline, gender, ethnicity and socioeconomic status.

Sensitivity analyses shows our results were broadly similar when varying definitions of hot/cold weeks were used (see Table S2 in the Supporting Information). The relationship between flares and hot weeks was strongest when temperatures > 95th percentile were considered ‘hot’.

Subgroups

To explore if the effect of weather differed according to the emollient type the children were randomized to at baseline, we investigated temperature–emollient interactions. The likelihood ratio test showed no evidence of effect modification (P = 0.55). A comparable analysis of baseline eczema severity (measured using POEM) also showed no modification of the effect of weather (P = 0.53).

Discussion

Our study demonstrates a seasonal variation in eczema symptoms in a cohort of UK children, with higher symptom scores in the colder winter and lower scores in the warmer summer. We found a reduced odds of a flare in hot weather compared with temperate weather. There was weak evidence that cold weather was associated with increased odds of a flare. Neither allocated emollient nor baseline eczema severity modified the effect of temperature on odds of a flare.

Our study benefited from prospectively collected eczema symptom data, in a cohort larger than those previously studied in countries with similar weather. Our sample is more representative of children with eczema in the UK, improving generalizability; previous studies generally recruited children from secondary care settings with more severe eczema. We are also, to the best of our knowledge, the first to examine whether associations between eczema severity and temperature differed by allocated emollient.

The use of a validated outcome measure (MIC in POEM) as a proxy for defining flares is also a strength. There is no consensus on how to define a flare (22 different definitions were identified in 1 systematic review), which is reflected in epidemiological studies that have previously examined associations between weather and eczema.^14,18 The initiative advises that using either the Eczema Area and Severity Index or POEM measured regularly throughout a study for at least 3 months is a reasonable approach.¹⁹

This was an exploratory secondary analysis treating data from a trial as a cohort; the original study was not designed or powered to specifically answer our research question. We were limited to the data collected for the purpose of the trial and thus were unable to account for other factors that can cause eczema flares, such as pollen, central heating use and pollution. The use of central heating can create sharp differences between outdoor and indoor temperatures and humidity, which could play a confounding role in our analysis. Improvements seen with warmer temperature may be confounded by ultraviolet exposure, as light therapy is a recognized treatment for eczema.²⁰

Our findings are useful to describe the effect of seasonal weather over weeks, but cannot comment on shorter changes within hours or a few days. This is because POEM scores asked about symptoms over the previous week and we utilized HadISD temperatures (which provides a daily average) converted into 7-day averages. Our weather data could also not account for time spent out of the patients’ home area. Although temperature across the different areas of England was broadly similar, we cannot account for travel further afield.

Utilizing an overall measure of eczema symptoms prevented us from looking at the effect of weather on specific parts of the body. Exposed areas such as the face and hands may be more susceptible to changes in temperature than protected areas such as the torso.

Previous studies have mainly been from outside the UK and focused on demonstrating a link between long-term climate and eczema. Global cross-sectional studies show that eczema prevalence increases, as latitude increases (and temperatures decrease).⁴ This is corroborated by studies conducted in the USA, Taiwan and Spain looking at hotter and colder regions.^5,6,21 Although prevalence and disease severity are different domains, our primary finding that hot weather is associated with fewer flares is consistent with these broad climatic studies.

Our findings are also consistent with the few previous studies examining short-term weather effects. Vocks et al. studied 2106 patients being treated at a specialist clinic in the Swiss Alps, taking daily weather measurements and eczema measures over an average time of 34 days.²² Analysis showed itch intensity improved with higher temperatures and sunshine duration. More recently, a cohort study of 170 children in the city of Seoul (South Korea) took daily weather recordings and eczema scores, showing that for every 5 °C increase in temperature, there was a reduction in risk of eczema present that day.²³

There have been some studies that show that it may be a change in weather, rather than the absolute conditions, which effect eczema symptoms. Byremo et al. conducted a small randomized controlled trial involving 56 children, with 2 trial arms. Those sent to subtropical Gran Canary benefited from an improvement in eczema scores, quality of life index, reduction in steroid use and reduction in Staphylococcus aureus bacterial skin colonization compared with controls who remained in Norway.²⁴ Our findings are in the context of the relatively temperate weather of the UK. Given the evidence from studies like the above, we postulate that in countries outside of the UK with harsher extremes of hot and cold temperatures a greater effect would be seen.

There are likely multiple factors underlying the reasons why flares are less likely in hot weather. Multiple laboratory studies have demonstrated a positive relationship between lower temperatures and decreased skin hydration and transepidermal water loss.²⁵ It is suggested that as temperature drops, the drier air encourages evaporation of water from the skin surface, which can leave the skin more vulnerable to damage. Silverberg et al. have also demonstrated an association between eczema prevalence and central heating use, which is usually turned on in cold weather.⁵ The frequent changes between heated indoors and cold outdoors are postulated to contribute to skin barrier disruption.

Questionnaire studies suggest that more of those with severe eczema perceive extremes of temperature to be triggers of eczema, compared with those with mild disease.⁹ Our findings indicate that eczema severity does not influence temperature-related flares, but further research in this area is needed to substantiate this.

Guidelines, patient charities and advice guides advocate for the switch to ointments in the winter months as the emollient of choice.^26–30 There have been no studies comparing the effectiveness of different emollient types in cold weather. Many advise the use of ointments because of their increased thickness and hydrating properties. However, our study indicates that the type of emollient has no impact on temperature-related eczema flares.

In conclusion, future research should investigate the differences in body site where flares are experienced in different weather and consider the role of other factors including humidity, pollen count and pollution. Meanwhile, parents of children with eczema should be advised that a combination of factors contribute to a flare. We provide some evidence to show temperature changes may play a role, specifically that hot weather is protective against flares. Switching emollients in different weather states to try and prevent a flare is unlikely to be helpful.

Acknowledgements

One of the authors, Amanda Roberts, is an expert patient having run an eczema patient support group for many years and caring for both herself and her children who have eczema. She provided insights into her lived experience of dealing with eczema and factors that matter to patients on a daily basis.

Funding sources

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. The BEE study was funded by the NIHR Health Technology Assessment programme (reference 15/130/07).

Data availability

The authors will support any reasonable request for the data that underpins the findings of this study in line with the original BEE trial data sharing agreement.

Ethics statement

Ethical approval: this was not required for secondary analysis of the BEE trial dataset; this used baseline and eczema severity data from participants who had given informed written consent for their anonymized data to be used in subsequent future research. The original BEE trial was granted approval by the NHS REC (South West – Central Bristol Research Ethics Committee 17/SW/0089). Informed consent: the study used baseline and eczema severity data from participants who had given informed written consent for their anonymized data to be used in subsequent future research.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher’s website.

Author notes

Conflicts of interest: The authors declare they have no conflicts of interest.