The literature review allowed us to identify relevant links that were used to draw a distinction of studies based on the type of relationship between indoor temperature and health outcomes.
4.1. Analysis of Direct and Indirect Links between Indoor Temperature and Health Outcomes
Table 1 presents the studies addressing each one of the links described in
Figure 3 and summarises the identified energy efficiency-indoor temperature-health outcomes.
The most prevalent steps that establish the direct relation between indoor temperature and health outcomes are links 1 to 5. This means that most studies take into consideration the influence and interconnection with outdoor temperatures or climate (link 1). The higher relevance assigned to socioeconomic factors-indoor temperature and indoor temperature-health outcome (links 4 and 5) compared to other direct relations, namely building envelope-indoor temperature and HVAC system-indoor temperature (links 2 and 3) is clear.
The influence of outdoor temperatures on the indoor environment has been assessed in studies such as Uejio et al. [
39] and Osman et al. [
40]. Several studies such as [
41,
42,
43,
44] have stressed that indoor temperatures were more significantly related to a given health indicator than the outdoor temperature.
Amongst the reviewed studies, only one study [
27] measured the association between temperature and health outcome indirectly based on outdoor temperature (link 6), without directly monitoring indoor temperature. However, changes in indoor temperature and health outcomes from energy efficiency studies have also been considered when resorting to existing empirical datasets. Energy Follow Up Survey or the English Housing Survey are examples of databases that integrate indoor temperature and energy efficiency used by studies associated to different links (e.g., References [
27,
45,
46] from the outdoor temperature-health outcome; building envelope-health outcome and HVAC system-health outcome—links 6, 7 and 8). The studies in References [
46] and [
47] are connected on the other hand, as they are based on the same empirical fieldwork data. As for, Rodgers et al. [
22] their study adopts a retrospective approach to investigate if empirical improvements to housing standards could lead to better health in householders.
Conversely, as expected given the empirical nature of the studies, most of the cause-effect relationships assessed are based on link 5 (indoor temperature-health outcome association). Within this scope, fewer studies have considered building envelope aspects (link 2), though when considered, this link is often associated with energy efficiency upgrades. From these, only one study [
44] has accounted for the existence of air conditioning (link 3). Though currently not considered a widespread feature, some studies forecast the increase of its relevance in the context of climate change [
51]. A large majority of studies in HVAC system-indoor temperature (link 3) (e.g., References [
22,
27,
29,
46,
48]), is therefore associated with heating in contrast to cooling systems. Many studies address both building envelope/HVAC system-indoor temperature (link 2 and link 3) because studies often consider simultaneously multiple energy efficiency improvements, such as thermal insulation and upgrades to heating systems. Although the focal point of the present work is indoor temperature, building HVAC systems (link 3) also contemplates ventilation issues. It should be mentioned that of the abovementioned studies, only Armstrong et al. [
27] take into consideration how insulation improvements may change ventilation and indoor air quality along with changes in winter indoor temperatures.
The number of studies in socioeconomic factor-indoor temperature (link 4) is indicative that socioeconomic influence seems to be more accounted for than building envelope aspects (link 2). The difference between these two links could be attributed to the fact that studies with health as a focal area tend to account for sociodemographic while largely disregarding building envelope aspects. Whereas studies focusing on energy efficiency take into consideration building variables from the upgrades or alterations to building envelopes, such as wall insulation or double glazing. This issue is consistent with existing literature, which has emphasised the relevance of household characteristics for health outcomes and energy efficiency contexts and how it has often been overlooked and considered a drawback from a health perspective ([
34,
35,
36]).
Nevertheless, a significant amount of studies from link 2 are also featured in link 4, being associated to social housing or the lower income segment of the population, where the probability of occurrence of worst housing quality is higher [
22,
29,
47,
48]. Yet, the highlight goes to one study [
29] that assessed relevant issues for both energy and health, such as fuel poverty status, financial difficulties and stress, food security, social interaction, thermal satisfaction and self-reported housing conditions.
These results show the segmentation between these focal study areas and denote the need to consider both socioeconomic and building aspects, in order to promote a better understanding of the association between indoor temperatures and health outcomes to contribute towards identifying potential energy efficiency measures to improve the characterisation of indoor temperature-health outcome (link 5), building envelope–indoor temperature (link 2) and socioeconomic–indoor temperature (link 4). The consideration of both latter links and fields of knowledge is desirable and could contribute to shifting health sector’s perception regarding the need for energy efficiency measures for healthcare reasons. In this sense, Jonathan Wilson et al. [
52] claims that a shift towards a more opened and receptive attitude from health sector would require more empirical evidence.
This insight is also in keeping with a key challenge pointed out by Haines et al. [
53], which consists on the need for the public health sector to establish partnerships with other relevant areas (e.g., city planners) and various stakeholders (research institutions, governmental and non-governmental bodies, public and private), in order to provide decision makers with well-grounded research evidence.
Meanwhile, studies that circumvent in loco monitoring of indoor temperatures have been categorised in HVC system–health outcomes and socioeconomic–health outcomes (links 7 and 8), as favouring less direct relationships to health outcomes.
No identified studies featured the interconnection between building HVAC system and health outcomes (link 9). The low number and/or absence of studies from HVAC system-indoor temperature (links 3) and HVAC system-health outcome (link 9) might be related to the fact that in Europe, in contrast to the United States, there is not a widespread adoption of air conditioning in the residential building stock [
51].
However, a recent review by Willand et al. has also cautioned that the householder’s response may also undermine the outcome of residential energy efficiency interventions, among which limited technical knowledge to deal with energy efficiency measures is highlighted [
32]. The relevance of the impact of technical aspects such as filtration on residential energy use has been further explored by Alavy et al. [
54]. While the integration of human dynamics in the control of HVAC systems has been studied by Jung and Jazizadeh [
55].
Yet the use of air conditioning is also closely linked to socioeconomic status, namely to household income. The Howden-Chapman [
56] study on energy poverty and health emphasised the increased vulnerability of low-income households with elderly. This segment of the population spends a high share of their income on energy. They are also more likely to be hospitalised for respiratory and cardiovascular conditions [
56]. Furthermore, Xu and Chen concluded that low income households have fewer energy efficiency appliances and less access to energy efficiency programs and require tailored policy measures to make energy more affordable and accessible [
57].
The strong association between income and household energy may also play a relevant role in a household’s adaptation to climate change and the choice between air conditioning and thermal insulation choices. De Cian et al. [
58] found that the future adoption of thermal insulation might be more difficult given that the adoption of air conditioning is promoted by income, urbanisation and demographic trends.
These studies anticipate the relevance of the interconnection between socioeconomic variables (link 4) to building envelope/HVAC system/indoor temperature and health (links 2, 3 and 9) with climate change. They also reinforce the need for conceptual frameworks to consider a time scale in the assessment of the relationship between household energy efficiency, indoor temperature, and health outcomes.
4.2. Analysis of the Relevance of Climate Change Timeframe for Health and Energy Efficiency
Moreover, though cardiovascular and respiratory health outcomes seem to prevail in the cause-effect column, different conditions and biomarkers are specific to each study. A more detailed perspective of these aspects is provided below.
The assessment of cause-effect relations on
Table 1 has enabled us to emphasise the high number of studies that focus health outcomes from exposure to low indoor temperatures (12 out of 15 studies) in contrast to Uejio et al. [
39] and Loenhout et al. [
44] that feature outcomes related to exposure to high indoor temperatures (2 out of 15 studies). Only one study reviewed (Armstrong et al. [
27]) considers both cold indoor and overheating exposures. The geographical distribution of these studies and indoor temperature, focus on overheating (in orange) versus cold indoors (in blue) or both (in yellow), is illustrated in
Figure 4.
Armstrong et al. [
27] present a longer timeframe than all others, contemplating winter and summer seasons, that is aligned with the nature of the health indicators used. This timeframe (of 10 years) is consistent with climate change concerns, and this is the only study that has looked to assess the perception of householders regarding home energy efficiency and climate change.
In addition, it is noticeable that studies focused mainly on developed countries. Noteworthy is the considerable amount of research developed in the United Kingdom (UK), which might result from the available national datasets with empirical indoor temperature measurements, as emphasised by Huebner et al. [
45].
Even within developed countries, results show that there is a considerable lack of empirical research in countries that have been greatly affected by cold homes and excess winter mortality such as Portugal, Malta, Spain or Greece. A recent body of research has emphasised that these countries are at the top rate of excess winter mortality out of a total of 30 European countries [
59]. These and other mild climate counties have been targeted as experiencing unacceptably low indoor temperatures [
60,
61].
Another observation is the scarcity of studies featuring indoor temperature monitoring during summertime, as emphasised in
Figure 4. This field of research has been considered scarce, despite the growing interest and concern within academic and local communities, as well as overall society. This argument has been supported by recent studies developed within either the health or more energy efficiency and indoor temperature oriented scopes [
27,
29,
39].
Departing from previously identified cause-effects in
Table 1, it is also possible to establish that a greater number of studies features morbidity outcomes comparatively to mortality.
Figure 5 illustrates the relation between the number of studies by health outcomes and study timeframe (in years). It is possible to see that there is a large diversity in study period considered for morbidity outcomes, that constitute 87% the of total number of studies. Yet, mortality outcomes (13% of total studies) are only associated with studies with longer timeframes (≥5 years). Once more, this result is in accordance with prior studies [
27,
46].
Given that longer timeframes in the research of indoor temperatures are also compatible with climate change research, results might suggest that mortality and morbidity outcomes could be considered in the context of climate change pathway if studies consider very long-term implications. However, in the current review, only one study, Armstrong et al. [
27], has a time frame beyond 10 years.
From a householder perspective, health has been considered a more relevant issue for the implementation of energy efficiency measures than climate change [
27]. Despite this, there is not much research to understand the impacts on health on longer timeframes, compatible with climate change issues, as illustrated by
Figure 5. Therefore, it is possible to imply that currently, householders might be misinformed and unaware of the real impact of the exposure to inadequate indoor temperatures on health and the relevance of energy efficiency in the context of climate change.
Consequently, there is an opportunity to leverage on the interest of people for health and promote studies with longer timeframes, that may lead to a better understanding of the impact of heatwaves on indoor temperature and health outcomes. Studies with very long timeframes would also contribute to understanding which energy efficiency measures could help improve health outcomes while mitigating climate change.
This result is aligned with Rodgers et al. findings, that emphasised that some of the co-benefits of energy efficiency may not be immediately perceived and that currently there is a scarcity of existing research in these terms, of the lack of long-term period studies [
22].
Armstrong et al. also claim that small sample sizes of empirical indoor temperature monitoring and lack of pre and post-intervention monitoring, make it difficult to determine accurately the impact of energy efficiency on indoor temperature and upon health outcomes [
27]. This may have implications in appropriately conveying health co-benefits from energy efficiency to decision and policymakers or even local communities. The consideration of longer timeframes could be crucial to address these issues and provide the scientific community with a more accurate and reliable empirical database to study the impacts of climate change.
Thus, very long-term studies could contribute to increased public awareness about climate change and its impacts and inform, based on empirical data, policymakers towards best available energy efficient solutions to mitigate them.
These results are in line with Willand et al. that have emphasised the need to integrate health goals into low carbon energy transition as a crucial aspect to develop an effective strategy for the housing sector [
62].
4.3. Analysis of Health Outcomes by Study
Four main categories of health indicators have been identified as being related to mental health disorders, cardiovascular or respiratory conditions, or other health outcomes. A more detailed listing of health indicators, specific for each study, is provided in
Table 2.
The relationship between the identified health outcomes, energy efficiency pathways (from
Figure 1) and indoor temperature is summarised by the Sankey diagram in
Figure 6.
The Sankey diagram represents fluxes between nodes. These fluxes show the transition from one node to another, here displaying possible energy efficiency pathways, health outcomes and indoor temperature identified in the reviewed studies. It is possible to see the extreme nodes (EE pathways and indoor temperature) are intermediated by a set of common health outcome nodes. The fluxes in a Sankey diagram are representative of the relevance of each health outcome as well as of the associated energy efficiency pathway. This relevance is estimated by the number of studies addressing each of the nodes. For instance, amongst the health outcomes, the least focused category is that of mental health disorders, with a thinner flux. However, its increasing relevance is recognised, as it is mentioned in studies that address diverse health outcomes, such as nocturia health indicator directly assessed by Saeki et al. [
42] or sleep onset latency (SOL) or difficulty in falling asleep assessed by Saeki et al. [
41].
In contrast to mental health outcomes, respiratory and cardiovascular conditions were the second and third most assessed health indicators, with wider fluxes compared to mental health.
Within cardiovascular and respiratory nodes relevant chronic conditions have been reported. According to the WHO [
15], the top positions of deadliest NCD at the worldwide level are occupied by cardiovascular disorders (e.g., heart attack and stroke) in the first place followed by cancer in second place and respiratory conditions (e.g., asthma and COPD) and diabetes in the third and fourth places. These four major NCD disease groups have currently affected approximately 17.9 million and 3.9 million people annually, with cardiovascular and respiratory conditions, respectfully [
15].
Yet the node for “other health” indicators category seems to be most representative, from indoor temperature to health outcome and from health outcomes to efficiency pathways, as illustrated in
Figure 6. This result is indicative of both the diversity and complexity of direct assessment of health impacts. A few examples of the complex interconnection between different health outcome nodes are given below and are detailed in
Table 2.
Most of the indicators categorised as “other health indicators” seem to be interconnected to other categories, namely for cardiovascular health conditions. For example, hypertension, mean arterial or blood pressure, and platelet count have been either mentioned or used as different health biomarkers in studies that aim to associate low indoor temperatures to health biomarkers for cardiovascular conditions (e.g., References [
41,
43,
49,
50]). Whereas respiratory cases have been in terms of relevance the second most assessed health condition with the least amount links to “other health indicator” categories. Its relevance comes from being directly assessed in empirical studies (e.g., References [
33,
45,
51]).
Since exposure to inadequate indoor temperatures (too cold or too hot) may imply adverse health impacts, some studies have suggested that through the improvement of indoor temperatures, health gains for householders could be achieved (e.g., References [
41,
42,
43]). Therefore in
Figure 6, health outcomes departing from inadequate indoor temperature are related to warmth/cool EE pathway. However, the fluxes are not directly connected, given that the studies tend to focus on each of the extreme nodes. This lack of connection between studies reinforces, once more, the need for a more holistic approach to address the relationship between energy efficiency-health-indoor temperature that better supports policy-making for efficient and healthy households.
The improvement of indoor temperatures on existing building stock is often, as previously shown in References [
22,
27,
46,
47,
48], linked to building envelope aspects (link 2 and link 3), associated to the adoption of energy efficiency measures.
4.4. Analysis of Housing Energy Efficiency and Health
In this section, the relationship between indoor temperature and health outcomes linked to the building envelope is assessed.
A total of 6 out of the 15 studies in
Table 2 are household energy efficiency related. A more detailed examination of energy efficiency measures adopted or mentioned is provided in
Table 3.
Based on
Table 3, it is possible to establish that thermal insulation measures were featured in all studied energy efficiency interventions, with wall insulation being the most adopted one. Improvements in household appliances namely heating systems and to windows and doors were also widely implemented. Other energy efficiency measures are residual comparatively to the previous categories for the studies considered in this review.
Wall insulation and heating systems contributed to increase the indoor temperature but not always to decrease relative humidity (RH), as shown in
Table 3.
Although each energy efficiency alternatives contributed differently for the increase in indoor temperatures, overall increases—for all intervention (e.g., References [
27,
29])—have been small, on average below 1 °C. However, even this slight increase has contributed in certain studies, such as for Poortinga et al. [
29], to reduce the number of hours exposed to very low indoor temperatures (<18 °C or <16 °C). Osman et al. [
52] also claim the adoption of EE alternatives has contributed to have fewer hours with unwanted temperatures (<21 °C) in the living room of chronically ill patients. It is also noteworthy that the largest improvements in indoor temperatures have been reached in critical living spaces in the household, such as the living room and the bedroom, where people spend their daytime and night time.
Besides the indoor environment, adopted measures have contributed to improving housing quality and efficiency standards [
22,
48], namely by reducing energy costs and increasing affordability. Furthermore, combinations of multiple energy efficiency alternatives, such as cavity wall and loft insulation with condensing boiler for the heating system, have reached considerable reductions (11.2%) in gas demand [
27]. Taking into consideration information from
Table 2 and
Table 3, regarding health indicators, it is also possible to say that to a large extent, a large share of studies featuring energy efficiency measures tend to focus on NCD health outcomes, compared to studies without energy efficiency measures.
Table 4 displays the most adopted energy efficiency measures vs. the most assessed health outcomes and impact on indoor temperature. The impact scale definition was inspired by Rodgers et al. [
22] and adapted for this case. The impact level represents, for each energy efficiency alternative undertaken, if its adoption contributed to improve or worsen a given health outcome. It can range from low to high impact level, with low level (--) being indicative of an undesirable change like an increase in hospital admissions. Conversely, a high impact level (++) is associated with a desirable change in the health outcome, such as a decrease in hospital admissions or medical appointment. Both these conditions are associated with a level of significance, to the
p-value for each study. No change (nil impact level) in the health outcome implies no association with a specific EE alternative with non-significant p-value. Most health outcomes reported in
Table 4 have been assessed individually for each EE alternative, with emphasis for insulation and heating systems. However, some results have been reported aggregately, either as a combination of all health outcomes (e.g., Reference [
27]) or as resulting from all intervention (e.g., References [
29,
48]).
It should also be noted that some of these studies have resorted to proxies in order to establish health status. For instance, Rodgers et al. [
22] and Armstrong et al. [
27] have resorted to hospital health statistics to assess the impact on health services such as emergency hospital admissions for COPD, asthma and mental disorders as a proxy for health outcomes. Besides hospital admissions Viggers et al. [
47], also considers self-reported health status and days off school and work, as well as visits to a general practitioner’s office.
From all studies reviewed, only one reported a negative association while the other eleven associations reported improvements in health, though with different levels of impact. Yet, the results also show that a significant number of conducted studies where the association between health indicators, energy efficiency and indoor temperatures was inconclusive. This is particularly highlighted for mental health and for cardiovascular health indicators, given the lower number of studies comparatively to respiratory conditions, as illustrated in
Table 4.
Among energy efficiency alternatives, insulation measures have contributed most for the improvement of respiratory and cardiovascular conditions, followed by alterations to windows and doors.
This beneficial association between energy efficiency measures and identified NCD’s could denote an effective course of action or opportunity to tackle some of the challenges in the health sector. A more detailed account of key findings for each study is provided in
Table 5.
There are also reports of other health outcomes, though less assessed and mentioned than NCD’s. This is the case of injuries and falls, that have had significant improvements, translated into less emergency hospital admissions for elderly householders that received upgrade to windows and doors [
22]. These results seem to indicate that more vulnerable groups have greater sensitivity to indoor temperatures and that even small readjustments could lead to significant changes regarding temperature-related diseases. This is true particularly for the elderly population, that tends to spend more time indoors. Other previous reviews have also highlighted the potential health benefits associated with even small increases in temperatures [
26]. Less objective outcomes such as “wellbeing” and “improved social interaction” have also been associated with improved indoor temperature from energy efficiency measures and might contribute indirectly towards better psychological health [
29].