Environmental change and health risks in coastal Semarang, Indonesia: importance of local indigenous knowledge for strengthening adaptation policies

ABSTRACT

Climate and environmental change are currently in the forefront as global development issues due to their economic, societal and health impacts, and the complex ways these interact. Experience and perception play an important role regarding the overall nature and severity of the environmental change, as these shape coping behaviour. Therefore, more insight is needed into local health-related risk perception, knowledge and coping mechanisms. In Semarang, a low-lying coastal city in Indonesia, a concurrent mixed-methods study design was applied using a cross-sectional survey supplemented by six focus group discussions and eight semi-structured key informant interviews with village officers and health-care workers. Respondents exhibited high awareness of environmental change and impact on health, housing and livelihood, and acknowledged the role of human activity, yet insight in underlying disease mechanisms was low. Most coping mechanisms were mainly short-term due to financial limitations and the frequency and intensity of environmental events. Existing health educational programs should be expanded and incorporate local perception and knowledge, to optimise coping behaviour and strengthen population resilience to environmental change. Further research needs to focus on differences regarding perception and desired coping strategies between household members and individual vulnerability to environmental change.

KEYWORDS:

• Environmental change
• health
• risk perception
• coping mechanism
• Indonesia

This article is related to:

Asian city futures: research to help inform spatial form and health

Introduction

One of the most pertinent issues in global development, climate and environmental change engenders detrimental economic, societal and health impacts. These changes affect the social and environmental determinants of disease and health. Changes in temperature, precipitation and air quality interact in complex ways with pathogens, disease vectors and pollutants, to produce health hazards within rapidly changing socio-economic and ecological contexts (Mcmichael et al. 2015, Barrett et al. 2015). Decisions to adjust to natural hazards hinges on awareness and risk perception, resources, as well as an appraisal of alternative coping responses or protective measures (Grothmann and Reusswig 2006, Terpstra et al. 2009). The extent to which environmental change is perceived as threatening to health and livelihoods depends on exposure, knowledge and the capacity to cushion detrimental effects through purposive action. Therefore, insight in local risk perception and coping mechanisms regarding health hazards caused by environmental change is needed, in order to develop adequate policies to support the adaptation of affected vulnerable populations. Moreover, as Dolan and Walker (Dolan and Walker 2006) posit, gauging local-level capacity to cope with change requires community-level research. Taking into account local knowledge systems and interpretations of the environment, and framing scientific knowledge on global changes in the local context, will make it more relevant for community-level adaptation and governance. This study provides a situational analysis of how vulnerable populations perceive health-related environmental risks and how coping mechanisms are shaped by local indigenous knowledge and perceptions.

Exposure to environmental hazards

Under circumstances of environmental stress, health is directly affected by increasing risk of accidents, but also indirectly by expanding vector-borne diseases and jeopardised water quality and food security, which increases the risk of diarrhoea, cholera and poisoning (Haines et al. 2006, Mcmichael et al. 2006, Kistin et al. 2010). Heatstroke and dehydration – especially affecting infants, elderly and those suffering from respiratory, cardiovascular and renal disease – result from rising temperatures, amplified by the urban heat island effect (Shahmohamadi et al. 2011, Heaviside et al. 2017). Climate change can also affect the psychological well-being through trauma, the creation of an unsafe environment, and by disrupting communities (Doherty and Clayton 2011). In particular low-lying island states, coastal areas, river valleys and semi-arid regions are vulnerable, due to the risk of coastal inundation, flooding, landslides and droughts (Oppenheimer et al. 2014). Highly (over)populated areas, such as megacities in low- and middle-income countries, are especially prone to increasing health hazards due to urbanization, slums, traffic congestion and industrial pollution, with mounting pressure on housing, sanitation, infrastructure and health services (Watts et al. 2017). Deteriorating environments associated with health hazards may lead to an increased propensity for human mobility, although this differs by socio-economic status, vulnerability and risk-awareness. In many cases, poorer populations bear a greater burden of environmental hazards, as they lack robust housing and inhabit vulnerable places such as river banks or slums built against hill slopes. These populations are less resilient and have less capability to cope with environmental stress impacts as they have fewer resources and lack strong social support networks (Black et al. 2011, Mcmichael et al. 2015). Differentials in risk exposure leads to the differential burden of disease and health inequity (World Health Organization 2002, Levy and Patz 2015).

Kahlor et al. (2006) distinguish ‘personal’ and ‘impersonal’ risks. Personal risks are environmental hazards directly threatening personal health or livelihood resources (e.g. infectious disease, flooding), whereas impersonal risks do not have an immediate individual detrimental impact (e.g. global warming), but rather act on a distant societal level. Impersonal risks may, however, transform into a direct threat in the long run when environmental degradation disrupts food resources and productive assets. Whether climate/environmental change is deemed a personal or impersonal risk influences behaviour and adaptive response. (Kahlor et al. 2006).

Knowledge

Knowledge and perceptions are formed by personal experiences, socio-cultural values, cosmology, cultural narratives and social networks. They are usually shaped through associative and affective learning processes, rather than being based on scientific data and analysis (Whitmarsh 2008, Weber 2010, Chowdhury et al. 2012, Lujala et al. 2015, Hasan and Nursey-Bray 2017). Lack of access to scientific and technical information on biophysical and geophysical processes may translate into subjective risk perceptions and alternative understanding when exposed to complex, often intangible new risks as global warming (Beck 1992, 1995, Etkin and Ho 2007, Armah et al. 2015). Yet, alternative knowledge is a valuable resource that can foster self-reliance and adequate coping behaviour in the context of uncertainty and adversity. This is especially the case when rooted in direct exposure and perception of changes in nature, as it increases personal relevance and personal risk awareness (Whitmarsh 2008, Baillergeau and Duyvendak 2016, Rogers et al. 2017). It leads to utilization of ‘in-between strategies’, i.e. neither completely rational nor irrational, utilizing intuition and trust, but based on prior experience and non-scientific knowledge rather than focusing on exact cause and effects (Zinn 2008). Moreover, as argued by Finn et al. (2017), traditional ecological knowledge is indispensable to understand health and disease from a non-Western scientific perspective (Finn et al. 2017).

Adaptive capacity

The nature and severity of the impact of environmental change on health partially depends on vulnerability to environmental hazards and risk awareness of environmental change as a direct threat to livelihoods, assets and health (Dolan and Walker 2006, Chun 2014). Vulnerability is conceptualised as the outcome of interplay between exposure (environmental stimuli), sensitivity (degree of impact) and adaptive capacity (ability to recover from environmental change events) (Comrie 2007). Sensitivity and adaptive capacity to environmental stress depend on personal traits, perception and behaviour, pre-existing health status, income and occupation, physical environment and social context, including access to productive resources, health services and support networks (Woodward et al. 2000, Mcmichael et al. 2015). Within the context of the sustainable livelihood framework, Chun argues that access to assets determines ones’ vulnerability, resilience and adaptability to cope with, act on and defend themselves to (detrimental) environmental changes (Chun 2014). Measuring (access to) assets can help explain differentials in risk perception, vulnerability and responses to environmental stress among populations. For example, poorer households often bear a greater burden of environmental hazards, as they often inhabit vulnerable areas and lack robust housing and are less resilient due to lack of resources and social support networks (Bierbaum et al. 2009, Black et al. 2011).

Despite accumulating evidence on the impact of extreme climate and environmental events on health outcomes, limited knowledge exists regarding specific situational health-related risk perception and associated coping and adaptation behaviours. It is imperative to expand this knowledge to gain insight in how risk-perceptions are related to coping with environmental change. It is also needed in order to develop targeted protective strategies for at-risk populations (Canham et al. 2014, Chun 2014, Verner et al. 2016). To address this knowledge gap, this study provides evidence on households’ health-related risk perception and coping strategies in a coastal area highly prone to various environmental threats, namely the city of Semarang, on north coast of Java, Indonesia. Results can be used by policymakers to assess the adequacy of extant local-level coping mechanisms and develop mitigation and adaptation policies, involving health facilities, non-governmental organizations and local community groups, in support of populations’ health challenged by environmental stress events.

Methodology

Study context

Semarang had 1,584,906 inhabitants in 2014, with a population density of 23,000 per km² (Badan Pusat Statistik Kota Semarang, 2014, Mulyana et al. 2013). The largest part of the city is located on higher ground, but the highest population concentrations are found in the lowland coastal area (Hadipuro 2012). For decades, inhabitants of the coastal area have been exposed to environmental hazards as heavy rains, flooding, heat waves and land subsidence (Marfai et al. 2008, Marfai and King 2008). Upland deforestation and land conversion have increased downstream sedimentation and flooding (Paripurno 2006). This has led to health risks such as dengue haemorrhagic fever and diarrhoea, which are exacerbated by deficient sanitation and waste management systems in the city (Hadipuro 2012, Khausarani Minanda et al. 2013).

Data collection and ethical approval

Data collection was conducted from November 2015 to February 2016 in two coastal areas of Semarang: Tambak Lorok, an urban industrial subvillage of the village Tanjung Mas in subdistrict Semarang Utara, and the semi-urban village Mangunharjo in subdistrict Tugu.

Evidence regarding health-related risk perception and coping mechanisms was collected using a concurrent mixed methods study design, including quantitative and qualitative methods (Creswell 2003, Creswell and Plano Clark 2007). Focus group discussions and key informant interviews were used to supplement and corroborate survey findings. Research instruments were validated through pre-testing and by building on validated questionnaires and previous studies on environmental change, health and coping mechanisms (Haque et al. 2012, 2013, Statistics Indonesia (BPS) et al., 2013, Toan Do et al. 2014). All instruments were pre-tested in a sub-sample of 20 respondents in neighbouring villages, and revised to fit the population. Ethical approval was obtained from the Commission on Health Research Ethics of the Faculty of Public Health at Diponegoro University Semarang and approval from the Badan Kesatuan Bangsa dan Politik of Semarang (Unit of National Unity and Politics). Letters of approval were distributed to subdistrict officers. All participants gave approval through oral or written informed consent.

Quantitative methods

Sample size calculation

A semi-structured survey was administered by students of the faculty of Public Health of Diponegoro University of Semarang under supervision.¹ Area sample size was calculated using a 95% confidence interval and unknown population proportion, resulting in a sample size of 200 households. Both areas appeared fairly homogeneous after rapid appraisal; therefore, the selection of households was conducted through simple random sampling based on a (sub)village household listing obtained from village officers, proportionate to the size of the neighbourhood (Rukun Tetangga/RT) and community (Rukun Warga/RW) groups. If house numbers were unavailable, households were selected using intervals proportionate to neighbourhood and community size (i.e. systematic random sampling). The head of household was chosen as the main respondent for the household and substituted by another adult household member when absent.

General measures, assets and wealth

Respondents were asked about household members’ demographic characteristics and household’s human physical, financial, physical and social assets and natural resources, including; educational level, occupation, possession of house and land, building materials and social networks. Educational level was measured as the highest level of completed education, based on the Indonesian education system. Occupation categories were predefined based on expert opinion. Employment security was scored through ranking based on expert opinion; occupations with high employment security (e.g. civil servant), average employment security (e.g. shop owners) or low to no employment security (e.g. street hawkers or housekeepers). Possession of land/fishponds and goods (e.g. number of bicycles or mobile phones) were measured ‘yes’ or ‘no’. Measurement of type of house, toilet and building materials was based on local valuation; higher scores were attributed to house ownership (instead of renting) and more qualitative and sustainable materials.

A wealth index score was calculated to reflect each household’s socioeconomic status, based on possession of relevant physical and financial assets, human capital and natural resources,² based on Filmer and Pritchett (2001) and the Indonesia Demographic Health Survey wealth index (Filmer and Pritchett 2001, Statistics Indonesia (BPS) et al., 2013). A factor score (wealth score) was calculated using a principal component analysis using SPSS version 20. The Kaiser-Meyer-Olkin (KMO) value was 0.728 and Bartlett’s test was significant. Cronbach’s alpha reliability estimate for the composite variable was 0.718. Wealth scores per household were ranked from lowest (poorest) to highest (wealthiest) and in ascending order divided into quintiles. Following Chun’s vulnerability framework, we assessed the relation between socioeconomic status and coping mechanisms and the necessity to undertake any actions to enable coping (Chun 2014).

Knowledge, risk perception and coping mechanisms

Main respondents were asked to describe (experienced) environmental change risks and impacts, current and future risk perception, and coping mechanisms related to households’ health, housing and income. Familiarity with environmental change and its impacts was measured through yes/no questions and provision of examples if possible. Knowledge was measured as ‘being able to identify examples’, rather than to explain a (more scientifically based) concept of environmental change. Risk perception, experiences and expected future events were measured through open questions. Answers were categorised as subtopics per theme (i.e. health, home and personal property, income). Respondents were asked whether their household conducted coping mechanisms related to health, house, income or ‘other’ and whether the action to enable coping was necessary, using predefined examples when applicable. Data entry was conducted by students and checked for inconsistencies. Using SPSS version 20, answers were quantified with descriptive and frequency functions. Relations were analysed using basic statistical analyses (e.g. t-test) and non-parametric variants when variables were not normally distributed, and controlled for demographic characteristics.

Qualitative methods

To gain more in-depth understanding of environmental change, risk perception and coping mechanisms, six focus group discussions were conducted in each area with heads of the community groups, male and female villagers. Participants were reached through the community health clinic (Puskesmas) or head of the community group. Topics – based on the survey – included knowledge of environmental change (impacts), risk perception, general and health-specific coping mechanisms. Four key informant interviews were conducted in each district with two local village officers and two primary health-care providers, focusing on the role of community and governmental institutions to provide assistance to cope with environmental change, and the role of health-care providers to assist local inhabitants to cope with health impacts.³

Results

General characteristics of the research areas and respondents

From each research area, 100 respondents were included, with a response rate of 100% (Table 1). Seventy-five percent of all main survey respondents were female and spouse of the head of household. Mean age was 43 years and all respondents were above 18 years old. On average, male respondents were older than female respondents. Highest level of education completed among all main respondents was slightly higher than elementary school and with no significant differences between males and females. Primary occupations for women included (small) business or trader and 39% of female main respondents was a homemaker. Thirteen female homemakers gained income through secondary occupation (e.g. ambulant trader, cleaning), indicating that the majority relied on income through other household members. Frequent occupations for males included (small) business owner/trader, fisher or construction worker. Average income of female main survey respondents was 1.000.000 to 1.500.000 IDR per month (75–110 US$), significantly lower than males. Principal component analysis indicated that the research population was relatively homogeneous in terms of wealth indicators and socio-economic status, see Table 2.⁴

Table 1. Area and respondent characteristics

	Tambak Lorok	Mangunharjo
Research area
Community groups	5	5
Neighbourhood groups	38	29
Households	1360	1631
Inhabitants	8686	5679
Survey households
Households included	100	100
Average number of family members	4.4	4.7
Dependency ratio (%) †	30.7	44.9
Male/female ratio	0.3 (23/77)	0.3 (24/76)
Mean age of respondent (sd)	43.2 (11.4)	42.1 (12.0)
Relation to head of household (%)
Head of household	27.0	28.0
Spouse	64.0	65.0
Child	8.0	6.0
Grandchild	1.0	1.0
Educational level (%) ‡
No education	6.0	4.0
Less than elementary school	12.0	5.0
Elementary school	23.0	25.0
Junior high school	30.0	34.0
Senior high school	19.0	27.0
University, tertiary or higher vocational education	5.0	4.0
Don’t know	5.0	1.0
Primary occupation (%)
Private sector (swasta)	23.0	18.0
Entrepreneur (wiraswasta)	14.0	8.0
Trader	11.0	10.0
Fisherman	7.0	3.0
Farmer	0.0	2.0
Paid housekeeper	4.0	0.0
Unemployed	3.0	6.0
Housewife	33.0	45.0
Student	1.0	0.0
Retired	2.0	1.0
Other (construction worker)	2.0	7.0
Income per month in IDR‡
No income	39.0	49.0
< 500.000	6.0	5.0
500.000–1.000.000	19.0	11.0
1.000.000–1.500.000	10.0	10.0
1.500.000–2.000.000	18.0	10.0
≥2.000.000	7.0	13.0
Don’t know	1.0	2.0

† Dependent household members include; anyone ≤15 years assumed to be unemployed (Statistics Indonesia (BPS) et al., 2013), retirees, disabled people or those not working for any other reason.

‡ Exchange rate at the time of survey, 1.00.000 IDR = 72.36 US dollar (31 December 2015).

Table 2. Wealth scores

Wealth quintile †	Tambak Lorok (n = 100)	Mangunharjo (n = 100)	Combined
Lowest	−1.895 to −0.977	−1.497 to −0.657	−1.771 to −0.809
Second	−0.481	−0.295	−0.368
Medium	0.082	0.177	0.131
Fourth	0.827	0.657	0.751
Highest	4.308	6.207	6.207

† Wealth quintiles show the range of wealth scores among population quintiles. Lowest quintile represents 0% to 20% poorest people, highest the richest 20%.

Experience and perception of environmental change

When comparing dry season (May–September) now to five to ten years ago, survey respondents in both areas noticed an increase in temperature, heat intensity and droughts, while the amount and frequency of rainfall were strongly perceived as less (Figure 1). Perceived temperature change during rainy season (October to April) by survey participants of both areas shows an increase in comparison to 5–10 years ago, and the start of wet season is perceived to have delayed. Amount and frequency of rainfall during wet season was mainly perceived to have decreased over the years. Respondents in both areas noticed an increase in land subsidence (Figure 2).

Temperature and weather often change, we cannot predict the condition of nature anymore.

- Healthcare worker, Mangunharjo

Figure 1. Perceived changes in dry season

Figure 2. Perceived changes in wet season

Participants in the focus group discussions and key informant interviews mainly experienced an increase in temperature and decrease of the amount of rainfall, a shifted start of rainy season to December/January, protracted dry season and increasing unpredictability of weather (supplement S1 and S2). Flood (water overflow in higher areas caused by rainfall, flash floods and waterlogging), and rob (overflow of seawater onto land lower than sea level), were perceived to have increased over the last few decades (Nugroho et al. 2013).

Other impacts experienced by all participants included an increase in one or more pests and disease vectors (mosquitoes, cockroaches, flies and rodents), damaged housing or property and loss of income (mostly due to damaged fish ponds). Respondents also noted the negative influence of development locally and in higher areas (e.g. building development, railroads).

Main future expected environmental events included flood (35%), rob (20%), land subsidence (8%) and drought (7%). Anticipated impact of environmental events included impact on housing and property (57%; dirty or damage), health (14%; including dengue) and not being able to perform daily activities (6%) such as retrieving water due to drought. Six percent of all respondents explicitly mention that they do not anticipate any environmental events or impact in the future as they see no threat. Experienced and anticipated environmental events and risks were not related to any demographic characteristics of the main survey respondent. The extent to which impact is experienced however did depend on geographical location. Families living near rivers and shorelines experience more (impact of) flood and rob.

In the past, when there’s rainfall at the uphill area, there’s no flooding here. But now, when there’s raining there, even for only one hour, it must be flooding here

- Head of community group, Mangunharjo

Knowledge

Sixty-six survey participants in Tambak Lorok versus 56 in Mangunharjo provided examples of environmental change, mainly concrete events such as flood and changes in precipitation (see Table 3). Of all respondents, few mentioned more abstract aspects and manifestations of environmental and climate change such as ozone depletion, pollution and global warming (see supplements). In both areas, the majority of those familiar with environmental change were also familiar with its detrimental impacts, more than half mentioned health impacts. Other impacts experienced included stunting of fish, deteriorated environment (damaged property) and air or water pollution (Table 3). No inhabitants of Tambak Lorok mentioned the impact of environmental change on crops, due to the industrial and urban characteristics of the area. Main knowledge sources of environmental change (impacts) include own experiences and mass media.

Table 3. Knowledge of environmental change and impact

	Tambak Lorok (n = 100)	Mangunharjo (n = 100)
Knowledge of environmental change (% yes)	66.0	56.0
Flood	31.7	24.8
Rain and strong wind	28.8	21.5
Temperature change	22.1	28.1
Drought	5.8	15.7
Direct human changes to environment (tree felling, building development)	6.7	0.8
Other (changes in weather/seasons, land subsidence, rob)	4.8	9.1
Impact of environmental change (% yes) †	84.6	80.0
Health	60.3	56.4
Environment	28.2	34.6
Crops	0.0	11.5
Cattle or fish	7.7	10.3
Other (mosquitoes, air or water pollution, economy)	3.8	4.3

† Question only applied when respondents were familiar with environmental change.

No demographic characteristics of the survey respondents were significantly related to the knowledge level of environmental change (impact) or type of changes or impacts mentioned. Qualitative data showed that most respondents related regular changes in weather and seasons to ill-health, e.g. relating colder temperatures to flu. Some also directly linked specific health problems to environmental events, such as contact with floodwater causing skin disease. Yet, respondents generally did not recognise aetiology of disease, and presence of pests, disease vectors and other environmental events were not (rightly) linked to specific health problems. Insight in these relations appeared unrelated to the main survey respondent’s gender, educational level or occupation, but predominantly seems to be determined by own experiences and perception, access to media, social networks and geographical location. Noticeably, qualitative data demonstrates that interviewees who work for the municipality, health-care facilities or community were more likely to have broader knowledge about environmental change (causality and scientific definition; see supplement S1 and S2). Health-care workers identified a relation between disease vectors and (increasing) incidence of infectious diseases, such as presence of mosquitoes and dengue incidence, relation between water contaminated by rodent faeces and pollutants causing diarrhoea and skin disease. They also noted that the current knowledge level of environmental change and health of most coastal inhabitants is low. A majority of all participants showed awareness regarding influence of human actions as littering, deforestation and industrial activity on causing environmental change or worsening natural environmental events. Inadequate waste disposal and pollution were identified as problematic behaviours detrimental to health, especially in combination with frequent flooding.

Coping strategies

Sixty-five percent of surveyed households used health-coping strategies based on personal experience and knowledge, including changing consumption of foods and drinks (e.g. more fresh ingredients) and changing behaviour (e.g. staying inside during hottest time of day or during extreme weather events) (see Table 4). Other coping mechanisms used in both areas include use of mosquito nets and insect repellent, self-medication (sun block, traditional medicine), seeking treatment (traditional or non-traditional) and changing clothing, e.g. protective clothing to prevent sunburn.

Table 4. Domains and mechanisms of coping

	Tambak Lorok (n = 100)	Mangunharjo (n = 100)
Health (% yes)	65.0	65.0
Consumption of foods and drinks	40.0	21.6
Behaviour	23.5	18.9
Self-medication	14.1	10.8
Use of mosquito net or insect repellent	9.4	20.7
Clothing	5.9	8.1
Seeking treatment (traditional and non-traditional)	5.9	15.3
Other (keep house or environment clean, exercise more) †	1.2	4.5
House (% yes)	81.0	54.0
Raised floor	62.9	32.1
Strengthened walls	11.3	18.5
Changed roof	8.1	12.3
Changed building material	12.1	11.1
Improved sanitation	4.8	21.0
Other (keep house clean) †	0.8	4.9
Income (% yes)	7.0	3.0
Migration (% yes)	19.0	18.0
Migration because of environmental change	12.0	4.0
Health insurance (% yes)	59.7	73.2

† Respondents answered cleaning to cope with environmental change, with a primary focus on either health or home.

Almost 60% of all household members surveyed in Tambak Lorok and 73% in Mangunharjo had health insurance, which could indicate a risk-averse predisposition. Qualitative data indicated similar health coping mechanisms and mention the financial benefits of having health insurance or visiting public health-care facilities over private facilities. They tend to directly seek healthcare even with mild symptoms due to low healthcare costs. Health-care workers pointed out the influence of low (health) educational levels, resulting in inadequate health-coping strategies and limiting locals to carry out adequate disease preventive measures.

An important coping mechanism is house renovation. In Tambak Lorok, 81% of all households made changes to their house in response to (perceived) environmental hazards; 63% raised the floor, 12% changed building materials and 11% reinforced walls. Fewer households in Mangunharjo made changes to their house (54%), but conducted similar measures, except for improving sanitation (21% in Mangunharjo versus 5% in Tambak Lorok).

Seven survey respondents in Tambak Lorok mentioned changing sources of income (e.g. changed occupation, using other production assets) to cope with environmental change in comparison to 3% in Mangunharjo (Table 4). Male and female main respondents generally express similar experiences and coping strategies. However, how these issues are presented seem to differ; in the focus group discussions, female participants generally seemed to approach environmental hazards from a family-oriented perspective (focusing more on housing, health), while males were more socio-economically oriented (focusing more on occupation, income reliability; see supplement S1 and S2).

In each area, almost 20% of survey participants mentioned wanting to move, although only a small part explicitly indicated environmental hazards as a reason for migration (Table 4). In-depth interviews point to finances and occupational limitations as barriers to migration. Most respondents are comfortable with their current living situation regardless of environmental events, due to social networks, facilities (e.g. schools, shops) and livelihoods. Although we were unable to retrieve information regarding permanently migrated households previous to the survey, participants mention people are still moving to the area.

If we don’t change our occupation, we cannot eat, because our fish ponds are already damaged. But to change our livelihood, it’s not easy; if we want to be a trader, we don’t have capital. If we want to be a factory worker, no factory would accept us.

- Focus group participant, Mangunharjo

As to economic impact of conducting coping mechanisms, a majority conducted coping mechanisms and indicated having taken action to enable coping mechanisms (89% in Tambak Lorok and 64% in Mangunharjo), through financial measures such as generating additional income, pawning or selling possessions, or taking (additional) loans (Table 5). Noticeably, within this group 10% of all survey households mentioned to have taken action to enable coping strategies while not applying any coping mechanism(s) against environmental hazards (yet). Some households noted that they were unable to conduct coping mechanisms due to current shortcoming financial means regardless of extra financial measures already taken. Others mentioned to have taken preliminary action to enable coping strategies in the future, which might imply a proactive strategy in anticipation of future environmental risks and impacts. Five percent of all households did not employ coping strategies as they did not experience any threats. However, an additional 4% did mention to employ coping strategies regardless of not experiencing any environmental threats. This might imply that several households have already adapted to environmental change in the area, seeing any behaviour to deal with environmental hazards as a part of daily life. This was confirmed in our qualitative data, where participants in both areas pointed out to have become accustomed with changing environmental conditions, as environmental change and hazards have been increasingly present for several decades. Conducting coping mechanisms and necessity to take action to enable coping mechanisms did not specifically and significantly discriminate by demographic attribute of the main respondent or the household’s wealth score.

Environmental change has been occurring here for such a long time. People have gotten accustomed to this condition.

- Healthcare worker, Tambak Lorok

Table 5. Economic impact of coping mechanisms

	Tambak Lorok (n = 100)	Mangunharjo (n = 100)
Action taken to enable coping mechanism(s) (%)
Yes †	89.0	64.0
Work extra	68.0	76.1
Pawn possessions	3.2	6.0
Sale or trade of goods	4.3	4.5
Sale of livestock	0.0	0.0
Extra loan	24.5	13.4
No, enough money	7.0	22.0
No, no environmental threat	4.0	14.0

† Participants were asked for examples if answered ‘yes’ (multiple answers possible).

Discussion

We provide evidence regarding households’ health-related risk perception, knowledge and coping mechanisms in relation to environmental change in a low-lying coastal region in Indonesia, exemplary for areas considered as most vulnerable to the detrimental impact of environmental change (Oppenheimer et al. 2014). Quantitative results from a survey were supported by findings from focus group discussions and key informant interviews.

Similar to other coastal areas on Java and developing countries, our study shows that populations of coastal Semarang are exposed to and experience effects of climate and environmental change (Mcgranahan et al. 2007, Haque et al. 2012, Hasan and Nursey-Bray 2017). Seasons are perceived to have shifted and temperature, heat intensity, droughts and overall yearly rainfall have increased, consistent with climate and meteorological data of Siswanto and Supari (2015) and Mulyana et al. (2013) (Mulyana et al. 2013, Siswanto and Supari 2015). Similar to previous studies, in addition to mass media being a source of knowledge, mainly experiences have shaped the knowledge and perception of inhabitants of Tambak Lorok and Mangunharjo (Whitmarsh 2008, Weber 2010). People do not rely much on scientifically based expert knowledge in assessing risks and choice of adaptive responses, but rather depend on ‘bounded rationality’ based on former experience with hazards, economic opportunities and strong sense of place and belonging, despite exposure to modern environmental risks (Beck 1992, Zinn 2008). The extent to which environmental changes were appraised as clear threats in terms of Kahlor et al.’s (2006) ‘personal’ and ‘impersonal’ risks varied. Most experienced and anticipated threats included direct events such as flood, rob, land subsidence and drought, mostly impact housing, followed by property and health. Very few inhabitants were also familiar with a more abstract notion of climate change as ozone depletion, greenhouse gases and global warming (see supplement S1 and S2). These findings match the results of a study in Jakarta’s riverbank slums where inhabitants expressed concerns about air pollution posing an indirect threat to health – in addition to primary concerns regarding floods (van Voorst 2015).

Coping strategies applied by inhabitants of coastal Semarang concern housing and property, livelihood and health. Socio-economic, cultural and geographical factors play an important role in shaping the coping behaviour families in high-risk areas of Semarang. Similar to findings in Bangladesh, our study population copes with health impacts of environmental and climate change by applying own knowledge (e.g. improving healthy behaviour, self-medication) and seeking treatment (Haque et al. 2013). While a majority of the study population was familiar with environmental and climate change, concerns were raised by health-care workers regarding the quality and efficiency of current local health strategies. Our survey confirmed that due to lack of adequate health education and suboptimal insight regarding causality and underlying mechanisms of disease subsequent coping strategies might be inadequate and short-term. Awareness of (perceived) environmental and climate change and insight in causality and disease mechanisms among interviewed health-care workers and municipality employees was noticeably higher. Hathaway and Maibach (2018) confirm that health professionals from English and non-English countries generally show awareness of climate change being harmful to health. Yet they also note self-assessed knowledge to be low, and that the perceived need to learn more is high, while being aware of the necessity of additional resources (Hathaway and Maibach 2018).

As to coping strategies regarding housing and property, Marfai and Hizbaron (2011) confirm (repeated) the use of short-term physical adaptation strategies such as heightening the floor in comparable areas in coastal Semarang (Marfai and Hizbaron 2011). Moreover, our study confirms previous findings regarding habituation of the coastal population to the burden of environmental change (Harwitasari and van Ast 2011, Marfai and Hizbaron 2011). Although possibly households who did have better health, access to resources and therefore better-coping capacity have migrated, people are still moving to the areal. Most respondents are comfortable with their current living situation due to local employment opportunities, social networks and facilities. Migration is not seen as viable or preferred strategy (Hillmann and Spaan 2017), contradicting studies elsewhere (Brown 2007, Mcleman 2011, The Government Office for Science 2011).

The frequency and intensity of environmental events seem to necessitate prioritisation of short-term coping mechanisms, rather than allowing inhabitants to save for potentially more beneficial long-term strategies as migration or health prevention strategies such as improved sanitation and waste management systems. This is affirmed by the finding that a majority of our study population needed to take financial actions to enable coping strategies. Observed coping strategies also indicate a certain resignation to insecure circumstances, while acknowledging compensating socio-economic advantages of staying put. This reflects Javanese culture, associated with sedentary lifestyles, belief in fate and striving for harmony, where hazards such natural disasters have spiritual meaning. The role of human agency in shaping one’s fate is acknowledged, albeit bounded by social norms and economic opportunity (Mulder 1978, Grossman, 2006, Schlehe 2006).

To investigate gender and socio-economic disparities and possible inequity between subgroups within our research population, subgroup analyses of the survey data were conducted (Levy and Patz 2015, Running 2015, Singer 2018, Rao et al. 2019). However, analyses showed no significant relationship between demographic characteristics of main survey respondents and perception or knowledge. Household’s wealth was also not significantly related to whether households conducted coping mechanisms or needed to undertake action to finance/enable coping mechanisms. Subgroup analyses could not differentiate well possibly due to sampling strategies or homogeneity of the study population (e.g. similar household structures, socio-economic status). More importantly, our survey was conducted on household level and while employed coping strategies seemed based on joint household decisions, our study did not focus on specifying individual differences between household members, such as knowledge, perception, experiences and personal coping preferences. Nevertheless, it is valuable to know an individual’s position and its role regarding decision-making within a household, as shown in a study by Nastiti et al. (2017) where household hierarchy and multi-generational occupancy (e.g. being husband or parent) determined decision-making. Here, female household members with low socio-economic status were only allowed to choose between simple, low-cost strategies, and had to consult a male household member or spouse if more costly solutions were required (Nastiti et al. 2017). Several findings in our study point to interpersonal differences as well. Significant income differences exist between male and female main survey respondents, and interviewed health workers, governmental employees or community workers exhibit more knowledge regarding causality. Moreover, focus of coping mechanisms seem to differ between men and women (socio-economic versus family), although Handayani et al. (2018) found no significant differences in climate changes adaptation strategies applied in male- and female-headed households in Tanjung Mas, part of our research areas (Handayani et al. 2018). In the context of Chun’s vulnerability framework, when discrepancies between individuals’ sensitivity and adaptive capacity within households exist, this may translate to disparity regarding individual vulnerability to environmental change (Chun 2014).

Based on our findings, our recommendations include practical changes to improve local knowledge and coping capacity regarding environmental change and health. Our research shows that (apt knowledge about) environmental change and health, as well as waste management and sanitation are currently problematic issues in coastal Semarang. Knowledge gaps and alternative understandings existing between general public and experts concerning climate change impacts on health are also noted by other studies, and need to be addressed (Etkin and Ho 2007, Chowdhury et al. 2012, Armah et al. 2015). Health educational activities should, therefore, focus on prevention and improving general knowledge on the relation between environmental change and health. These can build on existing educational activities provided by healthcare facilities and community centres. It is pertinent to link this to local indigenous or ‘alternative’ knowledge, as this is highly valuable for decision-making and can boost resilience when facing uncertainty and adversity (Baillergeau and Duyvendak 2016, Finn et al. 2017). In order to do so, health professionals need to be provided with the right resources, including expertise, personnel and financing (Hathaway and Maibach 2018). Moreover, to address possible discrepancies regarding vulnerability to environmental change and potential inequity, further scrutiny of interpersonal (e.g. gender) factors is necessary to allow policymakers to target and prioritize most vulnerable populations. In addition, longitudinal studies are required to raise awareness among public health authorities and quantify true disease burden caused by environmental change, in view of the complexity and multi-causality of the environment-health nexus. In agreement with van Voorst (2015), a broad approach is necessary to gain an adequate understanding of health risk perception and responses, thereby taking into account a whole range of risks relating to poverty and marginalisation and their interconnections (van Voorst 2015). This will support the development of more targeted policies including health promotion, vulnerability assessments and improved disease surveillance (Wirawan 2010, Braks et al. 2014). Although our results are only representative for (semi-)urban Indonesian coastal communities in Semarang, findings can provide insights on health-related perception and coping strategies in other coastal areas of developing countries exposed to environmental hazards.

Conclusion

Inhabitants of coastal Semarang see that their environment is changing and experience the impacts of environmental change such as illness and economic losses partly due to frequent and high-impact environmental hazards such as floods and extreme weather. Environmental events are linked to human action and the aggravating role of deficient sanitation and waste management is acknowledged. Yet insight in causality and underlying disease mechanisms is suboptimal, resulting in inadequate health coping mechanisms. In contrast to predicted migration, our study shows habituation to environmental hazards, linked to cultural and financial factors, of which the latter in combination with the intensity and frequency of detrimental environmental events leads to reactive short-term coping strategies. Future research should focus on quantifying disease burden and interpersonal differences (within households) regarding (health) impacts and the ability to apply preferred coping strategies in terms of environmental change. In addition, existing educational programs should be expanded and incorporate local perception and knowledge, in order to raise awareness of environmental risks, health impacts, influence of human actions and sanitation and waste management systems, and adequate coping behaviour to strengthen population resilience and adaptation capacity to environmental change.

Author contributions

JvdH and ES designed the study, analyzed the data and wrote the paper. All authors were involved in the development of the questionnaire. JvdH, BB and YHD conducted focus group discussions and key informant interviews. BB and YHD translated the focus group discussions and key informant interviews. All reviewed drafts of the paper and approved the final article.

Ethical approval

Commission on Health Research Ethics of the Faculty of Public Health at Diponegoro University Semarang, Indonesia.

Acknowledgments

The authors want to thank dr. Hanifa M. Denny, BSPH, MPH, PhD (former Dean of the Faculty of Public Health, Diponegoro University) for support and facilitation, and mrs. Toeti Rahajoe and the students of the Faculty of Public Health, Diponegoro University Semarang, for assistance in conducting the surveys and data entry. We also express our gratitude to dr. Hans Groenewoud and dr. Rogier Donders of the Department for Health Evidence, Radboud University Medical Centre for their advice regarding the statistical analyses and to Dr. Leon Bijlmakers for comments on an earlier draft. Lastly, we would like to thank dr. Ton van Naerssen, associate member of the Nijmegen Centre for Border Research (NCBR) at Radboud University Nijmegen, the Netherlands for proof reading the final manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed here.

Additional information

Funding

This study is a side-project of the New Regional Formations (NRF) project, supported by the Volkswagen Foundation. We are grateful for support received for this side-project from Radboud University Medical Centre, Nijmegen, the Netherlands, and Diponegoro University, Semarang, Indonesia.

Notes on contributors

Julie van de Haterd

Julie van de Haterd, Msc, is currently a final-year medical student at the Radboud University in Nijmegen, the Netherlands. She received a master’s degree in Biomedical Sciences at the Radboud University with a specialisation in Health Technology Assessment, consultancy and international public health. She is interested in interfaces between health care and social issues.

Budiyono Budiyono

Dr Budiyono Budiyono is currently associate professor and Dean of the Faculty of Public Health, Diponegoro University Semarang, Indonesia. He is a researcher in the field of public health, focussing on disease and health problems related to environmental risk factors. He has published several articles related health and environmental factors based on research supported by national and foreign grants, and received professional training supported by WHO, UNICEF, Diponegoro University and the Ministry of Health and the Ministry of Research, Technology and Higher Education Republic of Indonesia.

Yusniar Hanani Darundiati

Dr Yusniar Hanani Darundiati is a lecturer and researcher in Faculty of Public Health, Diponegoro University, Indonesia, specialized in environmental health. Her research focusses on heavy metal and pesticide effects on children development and childbearing mothers. Her dissertation was about mercury misuse in artisanal gold mining in Wonogiri, Indonesia and the effect on children development in order to develop a model to prevent this impact.

Ernst Spaan

Ernst Spaan MA, PhD, is Assistant Professor International Public Health at the Radboud Institute for Health Sciences (RIHS), Radboud University Nijmegen, the Netherlands. His research interests concern population and development issues, labour migration, environmental change impacts on health and livelihoods, and health systems reform in developing countries.

Notes

1. Students received training, including introducing study concepts, safeguarding confidentiality, reporting, dealing with interferences and contingencies, through group discussions and survey field-testing.

2. Assets included education and employment security of head of household, savings, possession of land and goods (radio, computer, mobile phone, refrigerator, air conditioner, television, bicycle, motorcycle, car/truck, generator, boat, sewing machine and stove), and housing quality (type of toilet, building materials and number of rooms). Non-discriminating variables were excluded from the analysis.

3. All focus group discussions and key informant interviews were audio recorded with approval of the participants, transcribed verbatim, translated by a native speaker for thematic analysis and cross-checked by two researchers. Results were categorised as subtopics per theme (knowledge, risk perception, coping mechanisms) and compared to findings of the survey.

4. Twenty-two variables were included to calculate wealth scores and strong indicators of wealth were possession of refrigerator(s), mobile phone(s) and motorcycle(s), and a number of rooms. Possession of a boat was negatively related to wealth, similar to the wealth index of the Indonesian Demographic Health survey of 2012 (Statistics Indonesia (BPS) et al., 2013). This is likely caused by the relation between owning a boat and being a fisherman by profession, an occupation with unreliable income in our research area due to the negative effect of environmental pollution on growth and harvest of fish. Two households, each including one head of household, showed outlying wealth scores due to a combination of high income through primary occupation, receiving remittances from family members, and possession of more goods, land or fish ponds. Outliers did not distort results and were thus included (Table 2).