Radon Investigation in 650 Energy E ﬃ cient Dwellings in Western Switzerland: Impact of Energy Renovation and Building Characteristics

: As part of more stringent energy targets in Switzerland, we witness the appearance of new green-certiﬁed dwellings while many existing dwellings have undergone energy e ﬃ ciency measures. These measures have led to reduced energy consumption, but rarely consider their impact on indoor air quality. Consequently, such energy renovation actions can lead to an accumulation of radon in dwellings located in radon-prone areas at doses that can a ﬀ ect human health. This study compared the radon levels over 650 energy-e ﬃ cient dwellings in western Switzerland between green-certiﬁed (Minergie) and energy-renovated dwellings, and analyzed the building characteristics responsible of this accumulation. We found that the newly green-certiﬁed dwellings had signiﬁcantly lower radon level than energy-renovated, which were green- and non-green-certiﬁed houses (geometric mean 52, 87, and 105 Bq / m 3 , respectively). The new dwellings with integrated mechanical ventilation exhibited lower radon concentrations. Thermal retroﬁtting of windows, roofs, exterior walls, and ﬂoors were associated with a higher radon level. Compared to radon measurements prior to energy renovation, we found a 20% increase in radon levels. The results highlight the need to consider indoor air quality when addressing energy savings to avoid compromising occupants’ health, and are useful for enhancing the ventilation design and energy renovation procedures in dwellings. ﬃ cient between radon and the of on radon level.


Introduction
Radon ( 222 Rn) is a colorless and odorless radioactive gas that has been strongly linked to deleterious human health outcomes, specifically lung cancer [1][2][3][4]. It is the most important source of ionizing radiation among those that are of natural origin, as it constitutes the second cause of death by lung cancer after tobacco [5]. While the outdoor radon rarely reaches epidemiologically significant levels due to atmospheric dispersal and dilution, in enclosed environments such as residences, the level of radon can accumulate at levels as much as two orders of magnitude higher than outdoors in inadequately dwelling characteristics; and (3) to investigate the influence of thermal retrofitting on indoor radon level. Passive samplers were applied for the radon measurements, and questionnaire surveys were used to collect information about dwelling characteristics and thermal retrofitting. The results of this study could be used to better understand the radon levels in energy-efficient dwellings and, potentially, to interpret the associated health risks. The study is also useful for improving the accuracy of exposure assessment of indoor radon, and for developing improved energy renovation strategies in terms of radon control.

Study Sample and Approach
Passive sampling of radon was performed within the framework of the 'Mesqualair' project on indoor air quality in energy-efficient dwellings from January 2013 to March 2016 in the western part of Switzerland. The owners of energy-efficient dwellings were selected from a list provided by the Romand Minergie Agency and the Cantonal Energy Service Offices. A total of 650 gave their consent to take part in the study (shown in Figure 1). A radon dosimeter was sent by post to each participating dwelling, together with a step-by-step instruction of sampling procedure (detailed in Section 2.3), and questionnaire regarding building characteristics. After sampling, the dosimeters were sent back to the project team for analysis. Two radon measurement campaigns took place during winter 2013-2014 (93 homes) and 2014-2015 (557 homes) to complete the 650 dwellings, while 616 homeowners returned the responded questionnaire in total.

Characteristics of Selected Dwellings
Most of the participating dwellings were individual or semidetached houses and most were occupied by owners. Among the 650 radon-tested energy-efficient dwellings, Minergie labelled buildings (M) accounted for 37% (217), and the remaining 433 homes were part of the national energy renovation program (Programme Bâtiment) for buildings (R). It is noteworthy that out of the 217 M buildings, most were newly built (NM, 182) and only 35 were renovated (RM), illustrated in Figure 1.

Characteristics of Selected Dwellings
Most of the participating dwellings were individual or semidetached houses and most were occupied by owners. Among the 650 radon-tested energy-efficient dwellings, Minergie labelled buildings (M) accounted for 37% (217), and the remaining 433 homes were part of the national energy renovation program (Programme Bâtiment) for buildings (R). It is noteworthy that out of the 217 M buildings, most were newly built (NM, 182) and only 35 were renovated (RM), illustrated in Figure 1. building structure in more than 50% of the sampled dwellings. A larger proportion of NM and RM dwellings had no natural ground floors (floors directly adjacent to the natural ground) compared to R dwellings. The majority of both M and R dwellings had completely excavated or back-grounded basements (schemed in Figure S1). The NM and R dwellings shared a similar distribution of garage type, while RM dwellings had higher proportion of outdoor parking. Only 36% of the selected dwellings were equipped with mechanical ventilation systems, distributed across 167 NM dwellings, 28 RM dwellings, and only 15 R dwellings. A relatively higher percentage of NM and RM dwellings were located in low radon risk regions. We received 432 effective responses regarding thermal retrofitting during energy renovation. The thermal retrofit included replacement of windows to reduce the air infiltration, renovation of roofs, and retrofitting of floors and exterior walls to increase the thermal insulation of the building envelop. A majority of the dwellings experienced replacement of windows and renovation of roofs, while <50% got floors and exterior walls retrofitted. Only 27% of renovated dwellings had all the three types of thermal retrofitting, as we called full retrofit. Since the focus of this study is to investigate the impact of energy efficiency status, as well as energy renovation on indoor radon level, we did not acquire information regarding radon remediation actions in the involved dwellings via the questionnaire.

Radon Measurement
In January 2013 and 2014, a radon dosimeter was sent to each participating dwelling. The owners were asked to follow the step-by-step instruction to install the passive sampling dosimeter (Radtrak 2 , Sweden, three-month detection range: 15-25,000 Bq/m 3 ) at least 1.5 m above the ground, and away from windows and doors in a heated and regularly-occupied room at the closest floor of the dwelling from the ground. The sampling was performed over three months during the heating season, to allow a reliable representation of the average annual indoor radon concentration, as per ISO 11665-8 Standard [45]. During the sampling period, the occupants were asked to keep their living habits as usual, without touching or moving the dosimeter. After three-month collection, the dosimeters were sealed by the occupants and shipped back to the project team. We stored the dosimeters in a dry place protected from light, and organized the shipment to the laboratory of Landauer Nordic, Sweden within one month. The dosimeters were then analyzed following the ISO 11665-4 standard [46].

Statistical Analyses
The statistical analyses were performed using SPSS 21 software and customized coding in MATLAB R2014 software. The concentrations of indoor radon were log-normally distributed (seen in Figure S2). Therefore, the parametric t-test (number of categories k = 2) and analysis of variance (ANOVA) test (k > 2) were performed to test the relationship between the logarithmical transformed radon concentrations and the dwelling characteristics, and thermal retrofitting. Since the distribution of detected radon concentrations followed the lognormal pattern, the geometric mean (geo-mean) can better represent the mean value of radon concentrations in different categories for comparison. On the other hand, the median value is always important for statistics of a dataset, regardless of data distribution. Therefore, we considered both in the study.

Radon Concentration
The distribution of indoor radon concentrations in Swiss energy-efficient dwellings is shown in Figure 2. Across 650 dwellings, the median value of detected radon concentrations was 71 Bq/m 3 , while the geo-mean was 85 Bq/m 3 with a geometric standard deviation of 2.8. The results were similar to those collected in the radon database of the FOPH for Swiss buildings (median value of 87 Bq/m 3 [30]), where the difference was mainly attributed to the disparity in sampling amount, period, and geographic distribution between this study and the FOPH radon database. Compared to the maximum recommended value of 300 Bq/m 3 , by the WHO [5] and the FOPH [14], radon concentrations in only 11% of sampled dwellings exceeded the threshold. However, considering the reference value of 100 Bq/m 3 by the WHO [5], around 40% of dwellings failed to meet the reference value. Notably, the maximum detected indoor radon level reached as high as 4280 Bq/m 3 , which is more than 40 times higher than the reference value.   . We observed significant differences in the radon concentrations in the three types of dwellings. The Minergielabeled (M) dwellings had significantly lower radon concentrations compared to energy-renovated (R) dwellings (geo-mean 56 and 105 Bq/m 3 , respectively). Interestingly, we detected different radon levels even between new (NM) and renovated Minergie (RM) dwellings. Radon concentrations in NM dwellings were significantly lower than those in RM and R ones: the geo-mean radon of NM dwellings (52 Bq/m 3 ) were 40% less than that of RM ones (87 Bq/m 3 ), and were only half of that in R homes (105 Bq/m 3 ). The difference in radon levels of the RM and R dwellings was not significant (p = 0.302). Compared to the FOPH reference level (300 Bq/m 3 ), radon concentrations in around 3% of NM, 6% of RM, and 14% of R dwellings exceeded the limit value, while the proportions of levels beyond the reference threshold (100 Bq/m 3 ) across the three types of dwellings became 20%, 37%, and 44%, respectively. The results indicate that energy-renovated dwellings (either Minergie or non-Minergie) had generally higher indoor radon levels than newly built Minergie-certified dwellings. The results imply the importance of the thermal retrofitting on indoor radon concentrations, which is discussed in Section 3.3.  . We observed significant differences in the radon concentrations in the three types of dwellings. The Minergie-labeled (M) dwellings had significantly lower radon concentrations compared to energy-renovated (R) dwellings (geo-mean 56 and 105 Bq/m 3 , respectively). Interestingly, we detected different radon levels even between new (NM) and renovated Minergie (RM) dwellings. Radon concentrations in NM dwellings were significantly lower than those in RM and R ones: the geo-mean radon of NM dwellings (52 Bq/m 3 ) were 40% less than that of RM ones (87 Bq/m 3 ), and were only half of that in R homes (105 Bq/m 3 ). The difference in radon levels of the RM and R dwellings was not significant (p = 0.302). Compared to the FOPH reference level (300 Bq/m 3 ), radon concentrations in around 3% of NM, 6% of RM, and 14% of R dwellings exceeded the limit value, while the proportions of levels beyond the reference threshold (100 Bq/m 3 ) across the three types of dwellings became 20%, 37%, and 44%, respectively. The results indicate that energy-renovated dwellings (either Minergie or non-Minergie) had generally higher indoor radon levels than newly built Minergie-certified dwellings. The results imply the importance of the thermal retrofitting on indoor radon concentrations, which is discussed in Section 3.3.

Associations with Dwelling Characteristics
As shown in Figure 4, the year of dwelling construction was strongly associated with indoor radon concentration (p < 0.001). Older houses had higher radon levels relative to more recently built dwellings, which was in agreement with the findings reported elsewhere [24,25]. We observe a negative linear relationship between log-transformed radon concentration and built year of dwellings (β = −0.002, R 2 = 0.09, p < 0.001, Figure S3). With the increase in building age, the geo-mean indoor radon concentration elevated gradually from 51 Bq/m 3 in dwellings built in 2000-2015, to 150 Bq/m 3 in dwellings built before 1900. Considering the reference level, only 3% of houses built in 2000-2015 exceeded 300 Bq/m 3 , while the exceed-limit proportion increased to 20% in dwellings built before 1900. We hypothesize that elevated levels of radon in old dwellings come as a combined result of inadequate sealing of the lowest floor against the ground and enhancement of airtightness of the dwellings without adjusting for ventilation needs.

Associations with Dwelling Characteristics
As shown in Figure 4, the year of dwelling construction was strongly associated with indoor radon concentration (p < 0.001). Older houses had higher radon levels relative to more recently built dwellings, which was in agreement with the findings reported elsewhere [24,25]. We observe a negative linear relationship between log-transformed radon concentration and built year of dwellings (β = −0.002, R 2 = 0.09, p < 0.001, Figure S3). With the increase in building age, the geo-mean indoor radon concentration elevated gradually from 51 Bq/m 3 in dwellings built in 2000-2015, to 150 Bq/m 3 in dwellings built before 1900. Considering the reference level, only 3% of houses built in 2000-2015 exceeded 300 Bq/m 3 , while the exceed-limit proportion increased to 20% in dwellings built before 1900. We hypothesize that elevated levels of radon in old dwellings come as a combined result of inadequate sealing of the lowest floor against the ground and enhancement of airtightness of the dwellings without adjusting for ventilation needs. Another important variable associated with indoor level of radon was geographical location of dwellings, as shown in Figure 5. The geo-mean radon concentration in dwellings located in a high radon risk region was more than three times higher than in dwellings built in a low risk region (200 vs. 56 Bq/m 3 , p < 0.001). Houses located in a low radon risk region also had significantly lower indoor radon levels than those in a medium risk region (geo-mean, 56 vs. 69 Bq/m 3 , p < 0.05). Only 0.01% of dwellings in low-risk areas and 5% of dwellings in medium-risk areas failed to meet the recommended radon exposure value. In contrary, 32% of the dwellings situated in high-risk zones exceeded the limit. The richness of radon in soil of high radon risk region can lead to the higher indoor radon levels caused by soil-building foundation transfer of radon [47], indicating the high importance of preventions for radon control in dwellings located in high radon risk regions. Another important variable associated with indoor level of radon was geographical location of dwellings, as shown in Figure 5. The geo-mean radon concentration in dwellings located in a high radon risk region was more than three times higher than in dwellings built in a low risk region (200 vs. 56 Bq/m 3 , p < 0.001). Houses located in a low radon risk region also had significantly lower indoor radon levels than those in a medium risk region (geo-mean, 56 vs. 69 Bq/m 3 , p < 0.05). Only 0.01% of dwellings in low-risk areas and 5% of dwellings in medium-risk areas failed to meet the recommended radon exposure value. In contrary, 32% of the dwellings situated in high-risk zones exceeded the limit. The richness of radon in soil of high radon risk region can lead to the higher indoor radon levels caused by soil-building foundation transfer of radon [47], indicating the high importance of preventions for radon control in dwellings located in high radon risk regions. Indoor radon concentration is highly associated with building ventilation, which can dilute accumulated radon with outdoor air, and in some specific cases prevent the radon infiltration through pressurization [48]. Figure 6 demonstrates that the mechanical ventilation can have a profound effect on indoor radon. Relative to naturally ventilated residences, dwellings with mechanical ventilation systems had significantly lower radon concentrations (geo-mean, 58 vs. 105 Bq/m 3 , p < 0.01). Similar findings were reported by other studies as well [25]. The need for mechanical ventilation in controlling the indoor radon is a priority for dwellings located in high radon risk regions, as evidenced by the increased difference in radon concentrations between mechanically and naturally ventilated houses (geo-mean, 96 vs. 251 Bq/m 3 , p < 0.001). We also observed significant associations between indoor radon concentrations and other dwelling characteristics, including type of the ground, building structure, type of basement, and type of garage. Specifically, dwellings with natural ground floor exhibited higher indoor radon levels compared to the ones without (geo-mean, 100 vs. 77 Bq/m 3 , p < 0.01), shown in Table S1. The natural ground floor allowed higher radon infiltration from the natural ground indoors, as similarly reported by Diallo et al. [49] and Collignan et al. [25]. The geo-mean radon concentration in dwellings with Indoor radon concentration is highly associated with building ventilation, which can dilute accumulated radon with outdoor air, and in some specific cases prevent the radon infiltration through pressurization [48]. Figure 6 demonstrates that the mechanical ventilation can have a profound effect on indoor radon. Relative to naturally ventilated residences, dwellings with mechanical ventilation systems had significantly lower radon concentrations (geo-mean, 58 vs. 105 Bq/m 3 , p < 0.01). Similar findings were reported by other studies as well [25]. The need for mechanical ventilation in controlling the indoor radon is a priority for dwellings located in high radon risk regions, as evidenced by the increased difference in radon concentrations between mechanically and naturally ventilated houses (geo-mean, 96 vs. 251 Bq/m 3 , p < 0.001). Indoor radon concentration is highly associated with building ventilation, which can dilute accumulated radon with outdoor air, and in some specific cases prevent the radon infiltration through pressurization [48]. Figure 6 demonstrates that the mechanical ventilation can have a profound effect on indoor radon. Relative to naturally ventilated residences, dwellings with mechanical ventilation systems had significantly lower radon concentrations (geo-mean, 58 vs. 105 Bq/m 3 , p < 0.01). Similar findings were reported by other studies as well [25]. The need for mechanical ventilation in controlling the indoor radon is a priority for dwellings located in high radon risk regions, as evidenced by the increased difference in radon concentrations between mechanically and naturally ventilated houses (geo-mean, 96 vs. 251 Bq/m 3 , p < 0.001). We also observed significant associations between indoor radon concentrations and other dwelling characteristics, including type of the ground, building structure, type of basement, and type of garage. Specifically, dwellings with natural ground floor exhibited higher indoor radon levels compared to the ones without (geo-mean, 100 vs. 77 Bq/m 3 , p < 0.01), shown in Table S1. The natural ground floor allowed higher radon infiltration from the natural ground indoors, as similarly reported by Diallo et al. [49] and Collignan et al. [25]. The geo-mean radon concentration in dwellings with We also observed significant associations between indoor radon concentrations and other dwelling characteristics, including type of the ground, building structure, type of basement, and type of garage. Specifically, dwellings with natural ground floor exhibited higher indoor radon levels compared to the ones without (geo-mean, 100 vs. 77 Bq/m 3 , p < 0.01), shown in Table S1. The natural ground floor allowed higher radon infiltration from the natural ground indoors, as similarly reported by Diallo et al. [49] and Collignan et al. [25]. The geo-mean radon concentration in dwellings with wood structures was less than half of that in residences of masonry or mixed structures (Table S2). The significantly higher radon levels in houses with masonry and mixed structures can be attributed to lower air infiltration. With respect to the influence of basement type, dwellings with semi-excavated or back-grounded basements had significantly higher radon levels than houses with completely excavated basements, as shown in Table S3. A possible interpretation is that dwellings with back-grounded basements had a living space directly above the soil, where the radon test took place, unlike the dwellings with fully excavated basements. The construction of semi-excavated basements may also entail greater risk of radon infiltration, given the larger number of cutouts in the building envelope in contact with the ground. Moreover, the completely excavated basement can act as a buffer for radon transmission between the soil and the living space. We obtained the analogous results for the garage type: dwellings with garage in the basement had significantly lower radon concentrations (Table S4).
There is a clear link between dwelling characteristics and indoor radon concentrations, which may explain the lower radon levels in newly built dwellings compared to renovated ones. In addition to exogenous important factors such as geographical region of Switzerland, building construction features that led to reduced radon levels and that should be recommended include installation of mechanical ventilation or controlled natural ventilation, and building airtight ground floors, as we noticed that all the seven homes with extremely high radon levels (>2000 Bq/m 3 ) were renovated ones located in high radon risk region but without mechanical ventilation.

Influence of Thermal Retrofitting
Based on 432 collected responses about type of thermal retrofit during energy renovation, we analyzed their influence on indoor radon concentrations. As presented in Figure 7a, dwellings with replaced windows with a goal to minimize heat exchange with the exterior led to slightly higher indoor radon concentrations compared to houses without retrofitted windows (geo-mean, 105 vs. 99 Bq/m 3 , respectively, p = 0.69). Similarly, renovated roof elevated radon levels by 9%, from geo-mean 99 to 107 Bq/m 3 , though without statistical significance either (p = 0.47). Retrofitting the floors and exterior walls increased the geo-mean radon concentration significantly by 38 Bq/m 3 (p < 0.001). Altogether, the 118 dwellings that fully implemented all the thermal retrofitting strategies, had the geo-mean radon concentration of 131 Bq/m 3 , 50% higher than other residences that had undergone partial thermal retrofit.
The influence of thermal retrofitting can be further interpreted by comparing the radon concentrations before and after energy renovation. From the Swiss national radon database, we retrieved radon concentrations data from 60 dwellings prior to their retrofit, which were involved in the current campaign. By calculating the ratio of radon concentration prior and after the energy renovation, we found on average 20% increase in indoor radon levels caused by thermal retrofitting: the geo-mean value increased from 165 to 197 Bq/m 3 . However, as shown in Figure 8, this increase in radon levels was not statistically significant (p = 0.15). In some dwellings, the radon concentrations increased by as much as 4-8 times after thermal retrofitting. Owing to increase in airtightness of dwellings after thermal retrofit, the air exchange rate decreased, leading to elevated indoor radon concentrations. Similar findings were reported in other studies with radon [33,37] and other air pollutants, such as formaldehyde and volatile organic compounds [36,50]. The influence of thermal retrofit on radon concentration explains the relatively higher radon level in RM dwellings than NM ones. The influence of thermal retrofitting can be further interpreted by comparing the radon concentrations before and after energy renovation. From the Swiss national radon database, we retrieved radon concentrations data from 60 dwellings prior to their retrofit, which were involved in the current campaign. By calculating the ratio of radon concentration prior and after the energy renovation, we found on average 20% increase in indoor radon levels caused by thermal retrofitting: the geo-mean value increased from 165 to 197 Bq/m 3 . However, as shown in Figure 8, this increase in radon levels was not statistically significant (p = 0.15). In some dwellings, the radon concentrations increased by as much as 4-8 times after thermal retrofitting. Owing to increase in airtightness of dwellings after thermal retrofit, the air exchange rate decreased, leading to elevated indoor radon concentrations. Similar findings were reported in other studies with radon [33,37] and other air pollutants, such as formaldehyde and volatile organic compounds [36,50]. The influence of thermal retrofit on radon concentration explains the relatively higher radon level in RM dwellings than NM ones.

Implications
The strong association between the presence of the mechanical ventilation in dwellings and reduced radon concentrations highlights the importance of adequate ventilation in limiting indoor radon exposure. Therefore, in both new constructed and energy-renovated dwellings, attention should be given to effective ventilation design and operation for control of indoor radon. Nonetheless, as indicated by the relatively high radon level in RM dwellings, installation of mechanical ventilation was not enough to ensure low levels of radon. We identified that occupants living in RM dwellings were not aware of the necessity to operate the mechanical ventilation in homes: some of them have never switched on the system. Relatively high radon levels in naturally

Implications
The strong association between the presence of the mechanical ventilation in dwellings and reduced radon concentrations highlights the importance of adequate ventilation in limiting indoor radon exposure. Therefore, in both new constructed and energy-renovated dwellings, attention should be given to effective ventilation design and operation for control of indoor radon. Nonetheless, as indicated by the relatively high radon level in RM dwellings, installation of mechanical ventilation was not enough to ensure low levels of radon. We identified that occupants living in RM dwellings were not aware of the necessity to operate the mechanical ventilation in homes: some of them have never switched on the system. Relatively high radon levels in naturally ventilated homes imply the importance of raising awareness of residents about window opening behaviors, which should accompany building retrofitting actions or new constructions.
It is also strongly recommended to implement radon measurements prior to energy renovation in order to adjust the renovation plan to effectively control indoor radon exposure. Radon prevention technologies need to be applied in cases of high radon concern. The main action is to install a specific sub-slab drainage against radon in order to make a depressurization and to extract radon from the ground before it enters a dwelling, as well as to enhance indoor ventilation [51]. In summary, to capitalize on the potential co-benefits of thermal retrofit in reducing energy consumption and maintaining high level of indoor air quality, we encourage stakeholders to pay special attention to adapting retrofit design based on specific building conditions (such as building age, construction type, and geographical location).

Conclusions
This study investigated the radon level in 650 energy efficient dwellings in western Switzerland. We examined the influences of building characteristics and thermal retrofit in new (NM) and renovated (RM) green-certified Minergie dwellings and in energy-renovated noncertified dwellings (R). We observed 40% lower radon levels in Minergie-certified dwellings, but there was no statistically significant difference between renovated Minergie (RM) and energy-renovated (R) dwellings. Indoor radon concentration was higher in older houses, especially in those built with masonry or mixed structures, and natural ground floors. Dwellings situated in high radon risk regions were prone to elevated radon risks. Installation of mechanical ventilation and completely excavated basement contributed to reduced radon concentrations in the living spaces. Thermal retrofitting of windows, roofs, floors, and external walls increased indoor radon concentrations, likely owing to reduced air exchange through air leakage.
Our results indicate that energy renovation measures without attention to indoor environment can adversely influence the level of indoor radon. Alongside the aggressive energy efficiency initiatives in Swiss buildings, these efforts should be accompanied with measures to minimize radon infiltration indoors and to secure adequate ventilation. Radon prevention constructions should take place in specific conditions, particularly for dwellings located in radon-prone areas like Switzerland. Alongside minimizing radon penetration from the ground, the ventilation design should take into account provision of a sufficient amount of outdoor air to dilute indoor radon either by mechanical means or by controlled natural ventilation. Occupants should be informed of the importance of indoor radon control, including renovating their ground floors and ventilating more often, especially in winter seasons. The recommendations should become part of the Swiss building renovation strategies and green-certification programs.