Background: Eczema is an important condition as it affects 20% of children in the UK and is associated with significant morbidity for children and their families. Although some progress in understanding factors associated with the occurrence of eczema has been made, very little is known about factors associated with disease worsening. Most textbooks and review articles quote long lists of exacerbating factors but with very little scientific data to support them. Before I could begin to study this topic, I first had to define a disease flare in eczema, systematically review the literature on flare factors in eczema and review available outcome measures for eczema.
Objectives: The objectives of the main study described in this thesis were to assess the role of various environmental factors on the severity of eczema in a cohort of children with eczema.
1. In hot weather, the combination of heat, sweating and grass pollen precipitates increased severity in children with eczema in the UK.
2. In cold weather, the combination of cold weather, indoor aeroallergen exposure and reduced relative humidity from central heating lead to increased severity in children with eczema in the UK.
These first two hypotheses were informed by previous research which proposed "summer" and "winter" types of eczema.
3. Detergents (soap, shampoo) increase the propensity to disease flares triggered by other factors at all temperatures, but more in cold weather due to impaired skin barrier function.
4. UK children with filaggrin mutations are more prone to the effects of climatic factors such as cold and heat than individuals who are wild type for filaggrin.
5. Any combination of greater than or equal to three exposures at any time is associated with worsening of eczema. The exposures assessed included: dust, exposure to pets, shampoo, sweating, swimming, nylon clothing next to the skin and a change in mean temperature of more than 3°C from the previous weekly average.
Pilot study: 30 children with moderate to severe eczema aged 0 to 15 years participated in a panel study over a one month period in June 2003 in Cork, Ireland. This study involved daily completion of a paper diary recording eczema severity and exposures. Feasibility of a panel study design was assessed and associations between exposures and disease severity were analysed.
Main study: A prospective cohort study (n=60) of children aged up to 15 years with moderate to severe eczema was studied for between six and nine months with overlapping start dates to allow study of seasonal factors. Exposures studied included: temperature, relative humidity, sun exposure, sweating, clothing, cleansing products/ washing, outdoor pollen level, extent and nature of exposure to household pets, dusty environments and swimming. Children or their parents completed daily novel electronic diaries recording eczema severity and exposures. Portable dataloggers were used to record indoor temperature and relative humidity. External meteorological data was obtained from a local monitoring centre.
The primary outcome was a daily "bother" score and the secondary outcomes were daily "scratch" scores and flares of eczema. Autoregressive moving average models (ARMA) were used to model the impact of each exposure on eczema severity for each individual. Standard random effects meta-analysis techniques were used to pool estimated coefficients across participants. Heterogeneity of responses as detected using Chi-squared tests represented inter-individual variation. The body site-specificity of reactions was also examined as was the interaction between filaggrin mutations and disease worsening with exposures.
Pilot study: The pilot study highlighted the issue of drop outs and missing data during the study. 83% (n=25) returned the diaries at the end of the study period, and within these, recording of disease severity was good (97% complete). However, there was variability in recording of exposures (65% to 83% complete). Preliminary findings suggested a temporal association between eczema severity and heat (lag 0, i.e. the day of exposure, p=0.04), damp (lag day 2, p=0.03), sweating and stress (lag day 3, p=0.03 and p=0.02 respectively) and damp (lag day 4, p=0.001).
Primary outcome: "bother scores":
Increased disease severity was associated with direct contact with nylon clothing (pooled regression coefficient 0.23, 95% CI 0.03 to 0.43), increasing exposure to dust (pooled regression coefficient 0.53, 0.23 to 0.83), exposure to unfamiliar pets (pooled regression coefficient 0.22, 0.10 to 0.34), sweating (pooled regression coefficient 0.24, 0.09 to 0.39) and shampoo exposure (pooled regression coefficient 0.07, 0.01 to 0.13). The association between shampoo use and worsening of eczema was enhanced in cold weather (pooled regression coefficient 0.30, 0.04 to 0.57). Body site specificity was observed for the reactions to nylon clothing, which was greater on covered sites (trunk p=0.02, limbs p=0.03), reactions to wool clothing on truncal covered sites (p=0.03) but not limbs (p=0.62), while worsening of hand eczema was associated with exposure to pets (p<0.001). The only interaction with filaggrin mutations was observed for the 2282del4 mutation and worsening of eczema in summer. Significant heterogeneity of responses between individuals was observed for exposure to grass pollen and outdoor temperature. In regard to the final hypothesis, a combination of any three of seven likely variables was associated with worsening of eczema (pooled regression coefficient 0.41, 0.20 to 0.63).
Secondary outcome: "scratch" scores: Increased disease severity was seen associated with swimming (pooled regression coefficient 0.14,0.00 to 0.28), exposure to wool clothing (pooled regression coefficient 0.28, 0.11 to 0.45), sweating (pooled regression coefficient 0.15, 0.04 to 0.26), shampoo (pooled regression coefficient 0.07, 0.01 to 0.13), dust (pooled regression coefficient 0.36, 0.12 to 0.59) and high grass pollen levels (pooled regression coefficient 0.10, 0.01 to 0.73).
Secondary outcome: flares of eczema: Only swimming was clearly associated with worsening of eczema using this outcome measure (pooled regression coefficient 0.42, 0.05 to 0.80).
Conclusions: The following factors were shown to be associated with disease worsening in children with eczema in this UK study: clothing (wool and nylon), sweating, shampoo, swimming, dust, contact with unfamiliar pets and high grass pollen levels. Relative to the study hypotheses, the association between shampoo exposure and eczema worsening was shown to be increased in cold weather. There was also evidence showing an association between various combinations of exposures and disease worsening. There was insufficient evidence to support the other hypotheses tested in this study but this may be explained by low prevalence of these exposures. The implications of the findings of this study for clinical practice are that for the first time, it has been shown that shampoo exposure may be associated with eczema worsening and that this is more pronounced in cold weather. This study also suggests that worsening of eczema may be more complicated in that multiple exposures acting in concert may be associated with worsening of disease. Future research with increased participant numbers is required to specifically study possible gene-environment interactions with filaggrin mutations and their relevance in relation to disease flares and to look at shampoo formulations in relation to worsening of eczema.