Prevalence, Incidence and Predictive Factors for Hand Eczema in Young Adults

A Follow-Up Study

Arne Johannisson; Ann Pontén; Åke Svensson


BMC Dermatol. 2013;13(14) 

In This Article


Study Group

This is the 13 year prospective follow-up study of a cohort of pupils in upper secondary school, 16–19 years old at the baseline assessment, and consequently they were 29–32 years old at follow-up. In 1995, 2,572 pupils in the four secondary schools in Växjö completed a self-administrated questionnaire regarding hand eczema, the response rate was 98.6%. Växjö is a town in southern Sweden with approximately 70,000 inhabitants.[15,16] In 1995, 74% of 16 – 19 years-olds attended secondary school in the study area, which was consistent with the overall attendance rate in Sweden. The 13-year follow-up of this cohort was performed in 2008. At both occasions the questionnaire was mailed in springtime. Swedish personal identification numbers were used to get updated addresses from the Swedish Population Address Register (SPAR). Addresses were found for 2,403 of the original 2,572 participants (Figure 1); 169 were unreachable: 106 had personal identification numbers not matching the SPAR register, 35 had emigrated, 21 had moved without providing a forwarding address, five were deceased, and two were not traceable for reasons of secrecy.

Figure 1.

The flow-chart of the cohort.


In 1995 the questionnaire was based on the Toulihampi questionnaire.[17] The questionnaire in 2008 was based on the Nordic Occupational Skin Questionnaire 2002 (NOSQ-2002),.[18] The questions regarding hand eczema were almost the same in the two questionnaires and the answer alternatives were exactly the same. Some additional questions constructed by the investigators were included in the 2008 questionnaire (See Additional file 1).

Topics surveyed by the questionnaire were: hand eczema, childhood eczema, asthma and rhino-conjunctivitis, household size and family structure, occupation and everyday activities, hand washing and skin care.

Distribution of the Questionnaire

A self-administrated postal questionnaire and a pre-paid return envelope were distributed in late May 2008. A postcard was sent at the beginning of June as a first reminder. At the end of August, a second reminder was sent which included a copy of the questionnaire, a pencil and a pre-paid return envelope. Finally, a postcard was sent in the middle of September as a third and final reminder.

Data Analysis and Statistics

One-year prevalence of hand eczema was estimated from reported hand eczema at present or having had hand eczema some time during the last 12 months (See Additional file 1). The question regarding the 1-year prevalence was previously validated.[19,20] The question on point prevalence was validated, and sensitivity (73%) and specificity (99%) were calculated.[15] To estimate the true 1-year prevalence for this cohort, a calculation of the 1-year prevalence in relation to sensitivity and specificity was made by using the formula: P = (P * + (specificity–1))/(sensitivity + (specificity–1)). P is the estimated true 1-year prevalence in the population and P* is the 1-year prevalence in the sample.[5,15,21]

The cumulative incidence was calculated on the individuals reporting having 1-year prevalence or ever having had hand eczema 2008 minus those who had 1-year prevalence or ever had had hand eczema in 1995. The cumulative incidence is presented as the percentage of new cases of hand eczema in the cohort. Incidence rate is presented as new cases per 1000 person-years, i. e. the cumulative incidence/13 years × 1000.

Four groups were constructed with the intention to analyse risk factors and the development of hand eczema over time. The groups were constructed as follows: those who reported having a 1-year-prevalence in 1995 and in 2008 are in group HX9508, those who reported having a 1-year-prevalence in 1995 but not in 2008 are in group HX95, those who reported having a 1-year-prevalence in 2008 but not in 1995 are in group HX08, and those who reported that they never had hand eczema are in group NoHX.

The reliability over time of self-reported childhood eczema in 1995 and then reporting the same in 2008 was determined by calculating positive predictive value (PPV); i.e. the percentage positive agreement in 2008 among the yes-respondents from 1995. The negative predictive value (NPV); i.e. the agreement of no-answers in 1995 and 2008 was also calculated.

Potential exogenous risk factors for developing hand eczema such as household size, time required for household work, frequency of hand washing, skin protective habits, working hours outside home and leisure activities were investigated by dividing the cohort into two groups. The respondents who had 1-year prevalence of hand eczema 2008, i.e. the merged groups HX9508 and HX08, denominated the HX group, and the group that reported never having had hand eczema, the NoHX group. Furthermore, hand eczema was also studied in the two hand eczema groups separately regarding these factors.

Regarding occupation, the respondents were asked not only to tell their profession, but also to give information about work tasks.

The groups HX9508, HX95 and HX08 were compared to the group NoHX using a multinominal logistic regression model. The endogenous factors childhood eczema, asthma and rhino-conjunctivitis as reported in 2008 were used. The response choices in this calculation were Yes/No. Exogenous factors such as hand-washing (times a day), usage of moisturisers (dichotomised Daily/Some time each week, some time each month, never), cooking, cleaning/washing laundry, and taking care of children 0–4 years of age (hours a day) were investigated.

Categorical data were presented as numbers and/or proportions in groups; quantitative data were presented by mean, median and quartiles. Nominal data were tested with the Chi-squared test. When the number of expected values was insufficient, Fisher's exact test was used. When comparing groups over time, McNemar's test was used. Ordinal and interval data were tested with Kruskal-Wallis H test and Mann–Whitney U-test in independent group comparisons. In the multinominal logistic regression analysis odds-ratios, 95% confidence intervals and p-values were given for all the covariates. If data was missing for any covariate, the individual was not included in the analysis. A p-value <0.05 was considered significant in all calculations. All statistical analyses were performed with SPSS 20.0 for Windows.


The study was approved by the The Regional Ethical Review Board in Lund, (application no 156/2008).