1 Introduction

Climate change is one of the most pressing issues of our time, with a range of associated effects threatening the health and lives of people, including sea level rise, global warming, weather events, and wildfires (IPCC 2022). In particular, climate change-induced heatwaves have significantly increased globally over the last decades, with a 68% rise in heat-related deaths among people over 65 years from 2000 to 2021 (Romanello et al. 2022). A recent review of the literature indicates that high temperatures are consistently associated with a higher risk of mortality and morbidity, particularly from cardiovascular and respiratory diseases (Conti et al. 2022). Several recent heatwaves have resulted in large numbers of deaths. The 2003 European heatwave resulted in 70,000 deaths (Robine et al. 2008), the 2010 Eastern European heatwave resulted in 55,000 deaths in Russia (Barriopedro et al. 2011), and 2500 died in the 2015 India heatwave (Mazdiyasni et al. 2017).

Developing countries are particularly susceptible to heatwave impacts, partly due to inadequate infrastructure and poor adaptation measures. This is especially true in South Asia, where more intense heatwaves, urban heat islands, and heat stress conditions are expected in the near future. According to S. Ullah et al., (2022), an estimated 185 to 492 million people will be exposed to daytime heatwaves, and 204–555 million to nighttime heatwaves in South Asia by the end of the century.

Pakistan is considered one of the most susceptible countries in the world to the effects of heatwaves, which are common from May to September and can exceed 50 °C (Rana et al. 2022). The 2015 heatwave in Karachi alone caused over 1200 deaths and 65,000 heat-related illnesses (Chaudhry et al. 2015). Within the Pakistani context, a heatwave is characterized by a scenario where the daily maximum temperature exceeds the average maximum temperature by 5 °C for a period of five or more consecutive days, with the reference period being 1961–1990 (Chaudhry et al. 2015). Moreover, the increasing trend of heatwaves is expected to be devastating to the country’s economy, as 40% of the labor force is involved in outdoor activities (FAO 2022).

Despite the risk posed by heatwaves in Pakistan, very limited research has been conducted to examine the social and cognitive factors that influence perception and behavioral responses during a heatwave (Bakhsh et al. 2018; Rauf et al. 2017). Similarly, Macktoom et al. (2023) highlighted the urban heat challenges faced by the public in Karachi, exploring the politics of shade and its vital but uneven access among outdoor workers. Abdullah et al. (2022) reported on the compounded vulnerabilities of the urban poor to extreme heat and COVID-19. Anwar et al. (2022) studied the governance of urban heat in Karachi, stressing the need for a holistic approach involving both state and non-state actors, while Abdullah and Macktoom, (2022) analyzed the concept of thermos-politics of urban heat in Karachi and highlighted the complex interplay of responses to heat stress. Growing evidence suggests the importance of implementing adaptive strategies against heatwaves, such as reducing physical activity, increasing water intake, finding cooling shelters, and staying in air-conditioned environments (Alcoforado et al. 2015; Howe et al. 2023).

However, to facilitate adaptation it is essential to thoroughly understand the processes and motivation behind the decisions of people to act or not. Studies on heatwave adaptative behaviors have highlighted different influencing factors such as risk perception, perceived self-efficacy, past experiences, socio-economic norms/beliefs, being old or being female, income, education, and social network among others (Akompab et al. 2013; Esplin et al. 2019; Khare et al. 2015; Rauf et al. 2017). An examination of current adaptative strategies used at a societal level, and an analysis of the challenges and facilitators can provide insights for future planning to effectively adapt to the increasing threat of heatwaves, especially in a heatwave-prone country like Pakistan.

The importance of experience likely depends on factors such as location, intensity, duration, type, and how it's defined (Hau et al. 2010; Ruddell et al. 2012). For climate change-related hazards, personal experience has often been limited to witnessing a specific event or suffering substantial losses from it (Sharma and Patt 2012). Socio-cognitive theories are increasingly being used to explore people’s perceptions and encourage adaptation to climate change-induced risks since knowledge and perception can motivate behavioral change. In this regard, the Health Belief Model (HBM) has been proposed as a useful theoretical framework for studying adaptive behaviors during heatwaves (Akompab et al. 2013; Arsad et al. 2022; Beckmann et al. 2021; Grothmann et al. 2017; Rauf et al. 2017; Richard et al. 2011; Valois et al. 2020; Wang et al. 2022). Briefly, this model postulates that individuals’ desire to take action to protect their health is contingent upon their belief that they are vulnerable to a particular risk, that the resultant consequences would be severe, that they have available options to prevent it, and that the benefits overweight the costs (Green et al. 2020). Therefore, the objective of the present study was to use the HBM to evaluate the factors associated with adopting adaptive behaviors to heatwaves in Pakistan. This study also explored determinants of adaptive behaviors during heatwaves in subgroups such as urban vs rural, heatwave experiences vs hypothetical scenarios, and unemployed, formal, and informal occupations. This approach enables us to comprehensively understand how individuals perceive heatwaves at different population levels and the adaptations they might consider. This study specifically focused on adaptive responses to heatwaves and not chronic heat conditions, with the questionnaire designed to capture heatwave-related perceptions and behaviors among a sample of volunteers registered with the Alkhidmat Foundation across Pakistan.

2 Methodology

2.1 Study design and respondents recruitment

A cross-sectional study design was used to assess heatwave adaptive behaviors across Pakistan. The study was carried out between February 2022 and September 2022. The researchers developed a questionnaire to collect data on participants' behaviors, and an online survey tool, SurveyMonkey (McDowall and Murphy 2018), was used to administer the questionnaire and collect responses.

The survey was distributed among the volunteers within the Alkhidmat Foundation—a non-governmental organization encompassing a diverse and multiethnic group of volunteers, working in nine regions and 150 out of 157 districts in Pakistan. The organization works in seven broad areas i.e., disaster management, health services, education programs, clean water, orphan care program, mawakhat program, and community services. The vast network of volunteers is particularly active in supporting the response to disasters in Pakistan. For instance, in the 2022 floods, more than 60,000 volunteers across the country were involved in rescue and relief activities in the affected communities (Alkhidmat Foundation 2022). The volunteers were recruited using a convenience sampling method reinforced by snowball sampling (De Moor 2022; Lake et al. 2019; Magnani et al. 2005; Mirzaei et al. 2020; Noga and Wolbring 2013; Staniford et al. 2011; Valerio et al. 2016). Through the use of social networks (WhatsApp groups, the organization’s messaging system) we aimed to maximize the number of respondents in this survey. Comprehensive messages in local languages such as Urdu and English were used. Participants of the study were encouraged to share the survey link with their fellow volunteers within the organization.

2.2 Questionnaire

The questionnaire was developed after a thorough literature review on heatwaves and the Health Belief Model (F. Ullah et al. 2024), piloted, and tested among 50 participants from Pakistan in both Urdu and English languages.

In its final version, the questionnaire was composed of four main sections. The knowledge about heatwaves was measured through 12 statements using a three-point Likert scale (Table S2). The Health Belief Model section encompassed the following constructs: perceived susceptibility (7 statements), perceived severity (7 statements), perceived benefits (6 statements), perceived barriers (4 statements), cues to actions (5 statements), and self-efficacy (5 statements) all using a five-point Likert scale (Table S3). The first two constructs of the HBM were used to measure risk perception, as previously proposed (Akompab et al. 2013). The adaptive behaviors section was assessed at two levels i.e., individual (14 statements), and household (16 statements) using a five-point Likert scale (Table S4 and Table S5). The last section collected data on the demographic profile of the participants. During the data collection phase, follow-up and reminder messages were sent out to the participants to enhance participation.

2.3 Data analysis

Analysis was conducted using Stata version 17 (Stata Corp. 4905 Lakeway College Station, TX, USA). Descriptive statistics (frequency distributions) and logistic regression were performed to identify significant relationships among the relevant variables. In our data analysis, we utilized three levels of significance: p < 0.01, p < 0.05, and p < 0.1, to evaluate the strength of our results with varying degrees of confidence.

Heatwave knowledge

In the case of knowledge about heatwaves, the “True” responses were assigned a value of “1” whereas the “False” and “I Don’t Know” responses were assigned a value of “0”. In the case of wrong statements, all the statements with responses as “False” were assigned a score of “1” and the “True” and “I Don’t Know” responses were given a score of “0”. The refined knowledge scores for all statements were summed to obtain a total knowledge score. The total knowledge score was then dichotomized as low (1–6) and high (7–12) (S1).

Heatwave perception

Risk perception was measured by multiplying the individual-level scores obtained for perceived vulnerability and perceived severity. Risk perception scores were then dichotomized into “low = 0 (49–500)” and “high = 1(501–1225)” (S1).

Heatwave adaptive behaviors

Adaptive behavior scores were dichotomized at all levels at the “midpoint”. For instance, the total individual adaptive behavior score was dichotomized into poor adaptive behavior (14–43) and good adaptive behavior (44–70). A third variable, overall adaptive behavior, was calculated by summing all scores obtained for individual and household adaptive behaviors by dichotomizing them into poor adaptive behavior (30–100) and good adaptive behavior (101–150) as shown in Table S1.

For the regression analysis, the dependent variable was adaptive behaviors in dichotomized form (poor = 0, good = 1), whereas the predictor variables were risk perception, perceived benefits, perceived barriers, cues to action, self-efficacy, heatwaves knowledge as a continuous variable, demographic information, and household characteristics as covariates. For all three levels of adaptive behaviors (individual, household, and overall), the crude and adjusted Odds Ratio with their 95% Confidence Intervals (95%CI) were estimated.

The study's purpose and confidentiality were mentioned at the beginning of the survey questionnaire. The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of the A.O.U. “Maggiore della Carità” di Novara (Protocol 1280/CE Study number CE 297/21) on 12th December 2021.

3 Results

3.1 Demographic profile of participants

The demographic profile of the study participants is reported in Table 1. Out of 1144 participants approached within the extensive network of registered volunteers with the Alkhidmat Foundation Pakistan, 698 participants completed the survey. Due to missing data, twelve questionnaires were removed from the final list of responses for data analysis, thus obtaining a final sample of 686. The majority of respondents were in the age group 35–54 (45.92%), followed by the age group 18–34 (29.01%) and 55 and above (25.07%). Most of the participants were male (71.14%) and married (71.57%), 71.14% of them had higher educational levels (Bachelor, Diploma, Master, etc.,) were employed (59.33%), and lived with others (91.11%). In terms of employment, the majority of the respondents were employed (59.33) and were engaged in formal occupations (48.10%). Ethnicity-wise, most of the participants were Pashtuns (34.99%), followed by Punjabis (17.78%), other ethnicities (17.20%), Urdu Speakers (16.47%), and Sindhi and Baluchi (13.56%). Province-wise, most of the participants were from Khyber Pakhtunkhwa (31.34%) followed by Punjab (25.66%), Sindh (17.06%), Baluchistan (11.37%), Azad Jammu Kashmir (6.27%), Islamabad (5.39%), and Gilgit Baltistan (2.92%). Most of the participants resided in urban areas (60.64%) compared to rural areas (39.36%). The analysis of this data revealed that the majority of the respondents (62.83%) confirmed experiencing a heatwave in their lifetime, highlighting their personal experiences with heatwaves.

Table 1 Characteristics of the Study Population
Full size table

3.2 Descriptive statistics of individual and household adaptive behaviors

At the individual level, the most common adaptive behaviors adopted by the participants during a heatwave were drinking a lot of water to stay hydrated (65.89%), seeking the protection of shady areas when out of home (46.94%), drinking more cold drinks such as milk, juices, kachi lassi, etc., (46.79%), increasing the number of baths per day (43.88%), and avoiding wearing dark color clothes when going outside (43.29%) (See Table S4). At the household level, the most reported adaptive behaviors were: restricting children from playing outdoors in a heatwave (63.99%), using fans, awnings, shades, and/or shutters, etc., (45.19%), using window shades to block strong sunlight and keep the home cool (44.31%), shut off the electronic devices when not in use (43.59%), and keeping an eye on ill, elderly people and babies in the family (43.29%). The study found that 79.30% of the participants had “good individual adaptive behaviors”, while 70.12% displayed “good household adaptive behaviors”. When combining individual and household adaptive behavior scores, the study showed that 66.47% of the participants had “good overall adaptive behaviors” (See Table S5). It is critical to bear in mind that the results stem from a study utilizing a sample from a local organization, necessitating extra caution in interpreting these findings.

3.3 Determinants of heatwaves adaptive behaviors

3.3.1 Determinants of individual adaptive behaviors

The univariable (crude model) analysis in Table 2 shows that participants with high-risk perception (OR = 1.484, 95% CI 1.024—2.150), high perceived benefits, (OR = 2.436, 99% CI 1.663—3.569), high perceived barriers (OR = 1.540, 95% CI 1.035—2.29), high cues to actions (OR = 1.954, 99% CI 1.321—2.892), and high self-efficacy (OR = 2.175, 99% CI 1.478—3.201) were more likely to have good individual adaptive behaviors during heatwaves. The association with high perceived benefits (OR = 2.048, 99% CI 1.327–3.159) and high self-efficacy (OR = 1.817, 99% CI 1.181–2.795) remained in the multivariable (adjusted model) analysis as well.

Table 2 Determinants of Individual Adaptive Behaviors
Full size table

3.3.2 Determinants of household adaptive behaviors

The univariable analysis in Table 3 suggests that participants with high-risk perception (OR = 1.633, 99% CI 1.175—2.271), high perceived benefits, (OR = 1.712, 99% CI 1.21—2.423), high perceived barriers (OR = 1.528, 95% CI 1.08—2.162), high cues to actions (OR = 2.103, 99% CI 1.474—2.999), and high self-efficacy (OR = 2.446, 99% CI 1.723—3.474), were more likely to have good household adaptive behaviors during heatwaves. The association with high cues to action (OR = 1.666, 95% CI 1.107—2.506) and high self-efficacy (OR = 2.133, 99% CI 1.444—3.149) remained in the multivariable analysis as well.

Table 3 Determinants of Household Adaptive Behaviors
Full size table

3.3.3 Determinants of overall adaptive behaviors

The univariable analysis in Table 4 highlights that those who had high-risk perception (OR = 1.369, 90% CI 0.994—1.884), high perceived benefits (OR = 1.906, 99% CI 1.358—2.675), high perceived barriers (OR = 1.697, 99% CI 1.211—2.379), high cues to action (OR = 2.555, 99% CI 1.803—3.622), and high self-efficacy (OR = 2.136, 99% CI 1.515—3.013) were more likely to have good overall adaptive behaviors during heatwaves. These results were substantially confirmed in the multivariable analysis (high perceived benefits OR = 1.444, 90% CI 0.979—2.13), high perceived barriers OR = 2.193, 95% CI 1.474—3.263) high cues to actions (OR = 2.193, 99% CI 1.474—3.263) high self-efficacy (OR = 1.743, 99% CI 1.187—2.560).

Table 4 Determinants of the Overall Adaptive Behaviors
Full size table

3.3.4 Determinants of adaptive behaviors: Subgroup analysis

In this study, we further stratified our analysis by three subgroups i.e., urban vs rural, heatwave experiences vs hypothetical scenarios, and unemployed, formal, and informal occupations. We divided the type of employment (occupation) into two categories according to the International Labour Organization (ILO) definition of formal and informal occupation (Ganzeboom 2010; Hussmanns 2004).

Urban vs rural

In urban areas, at the individual adaptive behavior level, only high perceived benefits (AOR = 1.676, 90% CI 0.949—2.961) were significant. At the household adaptive behavior level, high perceived benefits (AOR = 1.642, 90% CI 0.98—2.752) and high self-efficacy (AOR = 2.302, 99% CI 1.34—3.954) were found significant. For overall adaptive behavior level, high perceived benefits (AOR = 1.533, 90% CI 0.93—2.526), high cues to actions (AOR = 1.725, 95% 1.032—2.884), and high self-efficacy (AOR = 1.946, 95% CI 1.147—3.301) were significant predictors (Supplementary Table S6, S7, S8).

In rural areas, at the individual adaptive behavior level, high perceived benefits (AOR = 2.697, 90% CI 1.272—5.72), high perceived barriers (AOR = 2.098, 90% CI 0.889—4.95) and high self-efficacy (AOR = 3.164, 99% CI 1.512—6.62) were significant predictors. At the household adaptive behavior level, high perceived barriers (AOR = 2.346, 95% CI 1.156—4.758), high cues to actions (AOR = 2.969, 99% CI 1.471—5.996) and high self-efficacy (AOR = 1.944, 95% CI 1.028—3.675) were significant. At the overall adaptive behavior level, high perceived barriers (AOR = 2.901, 99% CI 1.44—5.846), high cues to action (AOR = 3.231, 99% CI 1.605—6.507), and high self-efficacy (AOR = 1.717, 90% CI 0.909—3.242) were significant predictors (Supplementary Table S9, S10, S11).

Heatwave experiences vs hypothetical scenarios

For respondents with heatwave experience, at the individual adaptive behavior level, high perceived benefits (AOR = 2.238, 95% CI 1.254—3.994) and high self-efficacy (AOR = 2.221, 95% CI 1.268—3.891) were significant. At the household adaptive behavior level, only high self-efficacy (2.286, 95% CI 1.386—3.771) was significant. For overall adaptive behaviors level, high perceived benefits (AOR = 1.641, 90% CI 0.996—2.704), high perceived barriers (AOR = 1.48, 90% CI 0.934—2.346), high cues to action (AOR = 1.804, 95% CI 1.107—2.94) and high self-efficacy (AOR = 1.817, 99% CI 1.181—2.795) were significant predictors (Supplementary Table S12, S13, S14).

For respondents within hypothetical scenarios, at the individual adaptive behavior level, high heatwave risk perception (AOR = 2.408, 95% CI 1.139—5.092) and high perceived benefits (AOR = 2.462, 95% CI 1.133—5.346) were significant. At the household adaptive behavior level, high perceived barriers (AOR = 1.999, 90% CI 0.967—4.132), high cues to actions (AOR = 3.537, 99% CI 1.706—7.333), and high self-efficacy (AOR = 2.2, 95% 1.086—4.457) were significant predictors. For overall adaptive behavior level, high perceived barriers (AOR = 1.828, 90% CI 0.911—3.669) and high cues to actions (AOR = 4.2, 99% CI 2.094—8.425) were significant predictors (Supplementary Table S15, S16, S17).

Formal, informal occupation and unemployed

In terms of formal occupation, at the individual adaptive behavior levels, high perceived benefits (AOR = 1.799, 90% CI 0.931—3.478) and high self-efficacy (AOR = 2.083, 95% CI 1.072—4.051) were significant predictors. At the household adaptive behavior level, high perceived barriers (AOR = 1.714, 90% CI 0.95—3.093), high cues to actions (AOR = 1.804, 90% CI 0.98—3.32) and high self-efficacy (AOR = 1.904, 95% CI 1.024—3.537) were significant predictors. At the overall adaptive behavior level, high perceived barriers (AOR = 1.782, 95% CI 1.013—3.135), high cues to actions (AOR = 1.767, 90% CI 0.971—3.215) and high self-efficacy (AOR = 1.678, 90% CI 0.912—3.087) were significant predictors (Supplementary Table S18, S19, S20).

In terms of informal occupation, at all three levels i.e., individual, household, and overall adaptive behavior, none of the HBM constructs were found to be significant (Supplementary Table S21, S22, S23).

In terms of the unemployed subgroup, at the individual adaptive behavior level, high perceived benefits (AOR = 3.203, 99% CI 1.493—6.875) and high cues to actions (AOR = 2.354, 95% CI 1.114—4.974) were significant predictors. At the household adaptive behavior level, high perceived benefits (AOR = 2.190, 95% CI 1.110—4.321) and high self-efficacy (AOR = 4.040, 99% CI 2.087—7.820) were found significant. At the overall adaptive behavior level, high perceived benefits (AOR = 1.819, 90% CI 0.955—3.464), high cues to actions (AOR = 2.843, 99% CI 1.501—5.385), and high self-efficacy (AOR = 2.240, 99% CI 1.211—4.146) were found significant predictors (Supplementary Table S24, S25, S26).

4 Discussion

Heatwaves threaten many communities around the world. Lack of preparedness and poor adaptation, resulting in increasing deaths and illnesses, is concerning. This study provides important insights from a country highly exposed to climate extremes. To the best of our knowledge, this is the first study investigating adaptive behaviors in Pakistan.

4.1 Predicting heatwaves adaptive behaviors: Insights from the health belief model

People’s choices to adapt to climate change are highly determined by their perceived beliefs (van Valkengoed and Steg 2019). We did not observe any association between risk perception and adaptive behaviors. Conflicting literature exists on this aspect, with some studies confirming our results (Ban et al. 2019; Esplin et al. 2019; Lee et al. 2019; Li et al. 2017; Wang et al. 2022) and others (Akompab et al. 2013; Rauf et al. 2017; F. Ullah et al. 2023) reporting a strong association between risk perception and heatwave adaptation. Risk perception is influenced by many factors, and it can vary significantly among different populations and within populations. Poor and marginalized communities are disproportionally exposed to multiple pre-existing conditions, which may increase their perception of their risk levels and contribute to unequal distribution of risks. For many individuals, the perception of health risk tends to be higher than that of other types of risk. Risk perception is a multifaceted concept that can be influenced by various factors such as the immediacy and severity of the risk, the knowledge of the risk, attitudes, experiences, personality, emotional well-being, and the social and cultural context (Rana and Routray 2016; Saqib et al. 2016; F. Ullah et al. 2020). Therefore, researchers need to incorporate the concept of risk perception in the assessment of health behaviors and the development of behavioral change interventions. Risk perception is an essential predictor of public support for mitigation and adaptation strategies to reduce the impacts of climate change-induced hazards such as heatwaves. In our study, the majority of the participants showed a high level of risk perception towards heatwaves confirming the findings of previous research (Akompab et al. 2013) and contradicting other findings (Rauf et al. 2017).

Our study found a significant association between perceived benefits and individual adaptive behaviors, corroborating previous research (Akompab et al. 2013; Andrade et al. 2019; Rauf et al. 2017). Except for the informal occupation subgroup, our study examined the role of perceived benefits in motivating adaptive behaviors across different contexts. Thus, interventions that increase perceived benefits may be effective in encouraging people to engage in healthy behaviors. This is also in agreement with the broader literature on climate change, which found that perceived benefits are positively associated with autonomous mitigation measures (Semenza et al. 2011).

Barriers such as lack of resources, lack of knowledge, institutional constraints, etc., can influence adaptive behaviors. Reducing heatwave catastrophic consequences requires that all stakeholders address the challenges and barriers to adaptation (Grothmann and Patt 2005). We noticed that perceived barriers had a significant effect on overall adaptive behaviors, suggesting stronger motivation to engage in adaptive behaviors contrary to the findings of past research (Rauf et al. 2017; Sheridan 2007). However, no significant link was observed between perceived barriers and heatwave adaptive behaviors at individual and household levels, confirming the findings of (Akompab et al. 2013; Andrade et al. 2019). Gerend et al., (2013) suggest that perceived barriers are multi-dimensional and differ systematically as a product of people’s behavioral intentions. Moreover, perceived barriers were significant predictors only in rural areas, among respondents with heatwave experience, hypothetical scenarios, and formal occupation. We note that although a positive association between perceived barriers and overall adaptive behaviors was observed in the current study, the latter was not measured directly. Therefore, it is crucial to interpret these results with caution when claiming an association between perceived barriers and adaptive behaviors. It must be said, however, that oftentimes perceiving high levels of barriers might be a driver for people to put more effort into engaging in behaviors that help reduce the impacts of different environmental stressors.

Cues to action such as access to information, past experiences, social influence, habits, etc., are the factors that motivate people to change behaviors. We found a significant link between cues to action and overall adaptive behaviors as well as household adaptive behaviors, consistent with prior research (Akompab et al. 2013; Rauf et al. 2017). This is in contrast with the finding of Andrade et al. (2019) which focused on health behaviors in response to other environmental threats. Furthermore, cues to action were found significant in urban and rural areas, within people who have experienced as well as those within hypothetical scenarios, formal occupation, and in the unemployed category. All these results could indicate that participants with more cues took more preventive behaviors during heatwaves because of their previous experiences.

To the best of our knowledge, the current study is the first study that incorporated the self-efficacy construct of the HBM to explore heatwave adaptive behaviors, as recommended by researchers (Akompab et al. 2013). Self-efficacy was a significant predictor of adaptive behaviors at all levels. Except for the informal occupation category, self-efficacy was found significant at all subgroup levels. Our findings confirm the findings related to other environmental threats (Andrade et al. 2019; van Valkengoed and Steg 2019) while contradicting a separate study (Baldwin et al. 2022) where self-efficacy was found to be lower among youth regarding taking action on climate change. Our findings indicate that people with higher self-efficacy are more likely to engage in adaptive behaviors because they believe they can adapt to changing environments. Interventions aimed at promoting adaptive behaviors should promote self-efficacy through measures such as access to information, awareness, and social support such as the availability of heat health advisories and toll-free helplines.

This study addressed adaptive behaviors both in the urban and rural contexts of Pakistan. The analysis revealed that while some HBM constructs were significant in both urban and rural communities, there were some differences between the two contexts. For example, in the urban context, high levels of cues to action and self-efficacy were found to be associated with good overall adaptive behaviors. Whereas in the rural context, high levels of perceived barriers and cues to action were linked with good overall adaptive behaviors. These findings suggest that these constructs play an important role in promoting heatwave adaptation in both urban and rural settings. Future research should focus on investigating predictors of heatwave adaptation in urban and rural communities using the HBM.

This study extended the results to those with heatwave experiences as well as hypothetical scenarios. For personal heatwave experience, perceived benefits, perceived barriers, cues to actions, and self-efficacy were all significant at the overall adaptive behavior level while in the case of a hypothetical heatwave scenario, risk perception and perceived benefits were found significant for the individual adaptive behaviors.

We also analyzed the results at different occupational levels to highlight any significant differences in the exposure to heatwave based on the types of occupation of the respondents. For example, perceived benefits and cues to actions were found to be significant in the unemployed category with no significant results observed in the formal and informal occupation categories. Overall, the results of the subgroup analyses did not differ much from the main analysis as highlighted above. While further research is warranted to fully operationalize all six constructs of the HBM and to thoroughly apply them to heatwaves, our findings can provide some useful inputs to policymakers and health specialists in designing effective interventions and targeted messages. This can be particularly valuable in Pakistan, which is currently developing its first-ever ‘Pakistan Cooling Action Plan (PCAP)’ as part of the Nationally Determined Contributions (NDC), slated for launch around 2026 (Ebrahim 2022). This can also be useful to tailor effective interventions to vulnerable subgroups of the population, such as outdoor workers, the elderly, athletes, and individuals with chronic diseases.

4.2 Future avenues for research

Future work should focus on employing larger and more diverse sample sizes to address the limitations of this study. Integrating other psychosocial theories with the HBM could provide deeper insights into how risk perception and associated factors influence adaptive behaviors during heatwaves. Research can also explore the psychological factors including cognitive biases, and emotional responses, that influence adaptive behaviors during heatwaves.

4.3 Study’s findings link with special issue ‘Climate change and wellbeing’

The study situates its findings within the narrative of ‘Climate Change and Well-being,’ highlighting the impact of climate change extremes like heatwaves on the well-being of the Global South’s populations. It explores the adaptive measures these communities undertake in response to the escalating heatwave incidences, emphasizing adaptation and the need for sustainable lifestyle and planning strategies that mitigate climate change’s effects while enhancing human well-being. Advocating for integrated policymaking, the research aligns climate adaptation and mitigation intending to improve living conditions, resonating with the global ambition to address climate change’s interconnectedness with human welfare, in line with the thematic essence of the special issue.

5 Strengths and limitations

This study has both strengths and limitations. Previous research in Pakistan had been limited to urban areas only (Rauf et al. 2017), while our study covered all the regions of Pakistan and obtained responses from both urban and rural participants. Moreover, the study examined adaptive behaviors at both the individual and household levels, yielding valuable insights from both perspectives. Among the possible limitations, it should be noticed that, as the analysis was not based on a random sample of the population, there could be a potential selection bias. The study's reliance and dependence on a small sample of 686 volunteers from an organization limits its reflections on Pakistan's broad demographic diversity, thereby hindering the universality of its findings, introducing considerable biases, and underscoring concerns about its wider applicability. We also acknowledge the fact that since the respondents are volunteers with the Alkhidmat Foundation Pakistan, there might be a possibility that they are more aware of different disasters occurring in Pakistan as they are usually active in the disaster response phase. Another limitation is the analysis of adaptive behaviors by incorporating responses from reported personal experiences and hypothetical scenarios. While self-reported experiences provide valuable insights, responses to hypothetical scenarios might not perfectly reflect actual behaviors during their real heatwave experiences. Future research should explore the use of additional data sources, such as real-time monitoring of actions taken during heatwaves as well as conduct research in areas with recurrent heatwaves across Pakistan, to further refine the understanding of adaptive behaviors. As the study was conducted over a long period, many of the responses will arguably be affected by incomplete/inaccurate memory or patchy retrospective recollections of the heatwave events, hence, future research should address this limitation. Another limitation includes relying on participants' self-reported classification of their residences as urban or rural, potentially introducing subjective biases. Likewise, as the study was conducted in Urdu and English, there is a possibility of linguistic bias because certain people might be more comfortable responding in other regional languages. The Urdu translation of the HBM itself might not perfectly capture the original HBM. The exclusion of the ‘Not Applicable (NA)’ in the Likert scale limits data for questions not universally relevant. For example, ‘wearing a hat when going outside’ might not apply to everyone. Without ‘NA’, participants might choose options that don’t reflect their reality, suggesting a need for more inclusive future survey designs. It is important to acknowledge that family members may not always be in agreement regarding responses to climate change as highlighted by Head et al., (2016). Therefore, it should be noted that responses on household adaptive measures might not necessarily represent the opinion of the whole family, as they often reflect the perspectives of individual family members who participated in the survey. However, it should be also noted that some of the associations were found both at the individual and household levels. Finally, we chose to dichotomize different variables in our analysis because of sample size constraints. Larger studies are warranted in the future to thoroughly evaluate the shape of the relationship between HBM constructs and adaptive measures using more flexible approaches.

6 Conclusion

Climate change is expected to change the world in the coming decades drastically. Heatwaves frequency, intensity, severity, and duration are predicted to increase in the coming years (IPCC 2022). As such, we need to prepare a resilient society that can meet the increased demand for adaptation during heatwaves. In light of the United Nations Secretary-General António Guterres, describing the latest IPCC report as no less than “a code red for humanity”, calls for decisive changes in our approach to the climate emergency are gaining momentum among scholars (Beckmann et al. 2021; Carman and Zint 2020; Noll et al. 2022; Valente et al. 2022). These calls are directed toward policymakers and decision-makers to engage and foster the implementation of individual and household adaptive behaviors. By examining responses in both actual and hypothetical heatwave scenarios, this research strengthens our understanding of how individuals and households assess heatwave risks and subsequently adopts to heatwaves. This is particularly relevant considering the future projections for South Asia, which suggest a rise in the frequency and intensity of both daytime and nighttime heatwaves (S. Ullah et al. 2022). No statistically significant relationship was observed between risk perception and adaptive behaviors, thus confirming previous findings (Akompab et al. 2013). The study highlights the usefulness of the HBM constructs for understanding heatwave adaptive behaviors. Perceived benefits (e.g., believing staying cool is helpful), and self-efficacy (e.g., confidence in taking cooling measures) strongly influenced individual actions during heatwaves. Along with self-efficacy, cues to action (e.g., warnings or instructions) were most important for household measures during heatwaves. By examining the effect of the constructs of the HBM on adaptive behaviors during heatwaves, this study enhances our understanding of how behavioral change theories can be applied effectively to assess people’s adaptative behaviors. Moreover, unlike other studies on heatwaves, this research applied all six constructs of the HBM, fulfilling a research gap previously identified by scholars. The findings of this study provide valuable insights into how individuals can be motivated to adopt certain behaviors during heatwaves. By leveraging the identified predictor variables, policymakers, disaster managers, health promotion specialists, and urban planners can all design and implement more effective heat health action plans to motivate behavior change during heatwaves.