Analysis of packaged drinking water use in Indonesia in the last decades: trends, socio-economic determinants, and safety aspect

Access to reliable, affordable, and safe drinking water is a human right and is essential to maintaining human health (Andualem et al., 2021; WHO & UNICEF, 2021). According to a report by the World Health Organization (WHO) and United Nations International Children's Emergency Fund (UNICEF) (WHO & UNICEF, 2021), the world is not on track to achieve the Sustainable Development Goal (SDG) 6 by 2030. Globally, there were still two billion people who did not have access to safely managed water services in 2020. Achieving universal coverage by 2030 in targets SDG 6.1 requires a quadrupling of current progress rates in safely managed drinking water services.

Globally, drinking water is obtained from various sources, including surface water, unprotected dug well or springs, piped water, boreholes or tube wells, protected dug wells or springs, rainwater, and packaged or delivered water (Rauf et al., 2015). Packaged drinking water (PDW) is one of the available drinking water options that is packaged in plastic, bottles, sachets, or bags in a range of sizes, and may be sold in shops, on the street, or delivered to homes (Fisher et al., 2015). PDW is rapidly growing in recent years worldwide. Global PDW consumption was predicted to increase by approximately 54% in 2017, i.e., approximately 391 billion liters, compared to 2007 (Statista, 2014). Global PDW consumption is estimated to increase by 513 billion liters in 2025 (Aslani et al., 2021). Factors such as health and cleanliness, the bottle, convenience, taste, and self-image trigger people to consume PDW (Ballantine et al., 2019). Furthermore, the safety aspect of PDW was another attractive aspect of PDW (Johnstone & Serret, 2012). A recent study in Cimahi, Indonesia, found that PDW, generally, has a good quality. The majority of the respondents consume PDW because they perceive better quality than other drinking water options and affordability for all income levels (Prayoga et al., 2021).

Studies found that PDW is not always safe although many people consider the safety of PDW. A previous study found that only 5% of bottled water purchased in Cleveland, USA, met the recommended concentration of fluoride, as recommended by Environmental Protection Agency (0.8–1.3 mg/L). Almost all tested bottled water samples met the fecal contamination standard (>1 of the colony-forming unit [CFU]/mL), whereas six samples contained 6–4.900 CFU/mL (Lalumandier & Ayers, 2000). A study in Nepal revealed that 6 of 24 samples of bottled water were fecally contaminated (Pant et al., 2016), while another study in one province in Indonesia revealed that approximately half of the sampled PDWs were fecally contaminated (Cronin et al., 2017). Unsafe PDW leads to waterborne illnesses and causes various health issues, e.g., diarrhea, malnutrition, and impaired physical and cognitive development (Cameron et al., 2021).

The use of PDW is increasing in Indonesia, but no study has further analyzed the trends and current status of PDW consumption in Indonesia. The current study aims to fill that gap. Furthermore, this study analyzes the socio-economic determinants of PDW consumption to determine the characteristics of PDW consumers in Indonesia. We complemented the analysis with a review of PDW quality from previous drinking water quality studies in Indonesia. These analyses are expected to provide a holistic view of PDW consumption in Indonesia and enable stakeholders to design relevant policies and strategies on this topic, especially in Indonesia. Thus, this study contributes to the analysis of drinking water-related practices in developing countries.

This study conducted three analyses, including (1) trends of PDW in the last two decades, (2) socio-economic determinants of PDW consumption, and (3) a systematic literature review (SLR) of PDW in Indonesia that focuses on the water quality. Further information about the step of analysis is provided in each section.

The trends in the use of PDW in Indonesia were obtained from the website of the Indonesian Statistics (BPS; https://www.bps.go.id). The collection of statistical data in the Indonesian National Socio-Economic Survey (SUSENAS) was conducted from selected households through face-to-face interviews. This survey is conducted twice a year, in March at the district/city level, and in September at the provincial level. The main results are then presented on the BPS website, i.e., percentages for national and provincial levels. The yearly data on PDW consumption from 2000 to 2020 was extracted and used in this study to see the trends of PDW consumers in Indonesia in the last two decades. Linear regression in Microsoft Excel was conducted to estimate the future PDW consumption in Indonesia.

Moreover, we were interested to confirm the hypothesis that PDW consumption in Indonesia is correlated with economic growth. In this study, the economic growth is represented by the urban population and gross domestic product (GDP) per capita. Bulut & Seçer (2019) found that population growth and urbanization influenced PDW consumption. Moreover, GDP annual growth rate has a significant positive correlation to PDW demand (Zhang et al., 2017). The urban population and GDP data were obtained from the World Bank data (https://data.worldbank.org/country/indonesia). The bivariate Pearson correlation in IBM Statistical Package for the Social Sciences (SPSS) version 26 was conducted to assess the correlation between them (IBM SPSS Statistics, 2021).

We used the Indonesian Demographic Health Survey (IDHS) data (https://dhsprogram.com/) from three periods, including 2007, 2012, and 2017. IDHS was a cross-sectional survey that collects data from a representative sample of the population in the countries that participate in The DHS Program. IDHS was conducted by Indonesian Statistics in collaboration with the National Population and Family Planning Board (BKKBN) and the Ministry of Health (National Population and Family Planning Board et al., 2018). These three periods were used to determine the pattern of socio-economic determinants of PDW consumption in Indonesia, i.e., whether or not there is a variation or change of determinants over time.

We combined information from household-level and women's individual-level data from the IDHS datasets to perform a statistical analysis of socio-economic determinants of the PDW consumption in Indonesia because some variables were located in household-level data only, and vice versa. Data on the source of drinking water, sex and age of the household head, type of residence, region, and type of toilet facilities are available at the household-level data. Meanwhile, data on the mother's educational level, father's educational level, and mass media exposure are available at the individual-level data. After combining these datasets, 29,840 respondents in DHS 2007, 31,173 respondents in DHS 2012, and 31,253 respondents in DHS 2017 were used in the statistical analysis.

The PDW variable was obtained from the drinking water source information at household-level surveys. It was coded as a binary variable, wherein households who used non-PDW were assigned as ‘0’ and those who used PDW as ‘1’. This study defined PDW as refilled and un-refilled water, which has been treated before being distributed or sold to the consumer.

This study selected the socio-economic determinants of the PDW consumption based on two criteria as follows: (1) previous studies on PDW or studies that analyze socio-economic factors that determine the types of drinking water sources in Indonesia and other developing countries (Table 1) and (2) the availability of variables in the IDHS datasets. Based on these criteria, the following nine variables were used in this study: sex of household head, age of household head, mother's educational level, father's educational level, wealth index, type of residence, region, type of toilet facility, and exposure to mass media. Table 1 shows the detailed categorization.

The variable exposure to mass media was created from three variables, namely, the frequency of reading newspapers or magazines, listening to the radio, and watching television. Each of these three variables was coded as a frequency value between 0 and 3 (0=not at all, 1=less than once a week, 2=at least once a week, 3=almost every day). These values were then summed to get a score that measures the exposure to mass media, with a minimum score of 0 and a maximum score of 9.

Other variables were directly obtained from the datasets and used without any significant modification. We used the wealth index categorization and the mother's educational level, which are available in the datasets.

The three IDHS datasets were collected and entered into Microsoft Excel for data cleaning and merging of an individual- and household-level data based on case identification. The merging of an individual- and household-level data was used because some variables were located in household-level data only, and vice versa (Table 1). Afterward, the data from Microsoft Excel were entered into SPSS version 26 for statistical analyses. Logistic regression was used to find significant socio-economic determinants of PDW consumption in the three IDHS datasets, i.e., three regressions were separately conducted for each dataset. The independent variables include the socio-economic variables and the dependent variable includes PDW consumption.

The SLR aims to unveil the water quality of PDW in Indonesia. The preferred reporting items for systematic reviews and meta-analyses (PRISMA) were followed. PRISMA is a guideline to assist the researcher in conducting systematic reviews and meta-analyses by preparing the high standard protocol (Moher et al., 2015).

Studies collected for SLR were extracted from three indexed databases, including ScienceDirect, Scopus, and PubMed. Keywords used were as follows: ‘water quality’ and ‘commercial drinking water’ or ‘packaged drinking water’ or ‘potable drinking water’ or ‘refilled drinking water’ or ‘bottled drinking water’ and ‘fecal contamination’ or ‘Escherichia coli’ and ‘Indonesia’. The elimination of duplicated articles was also conducted in this phase.

The screening process is conducted by scanning the article by title and abstract to select and ensure the relevance of the article to the research topic. Furthermore, we selected articles that contain information on the water quality of PDW, either physical, chemical, or biological aspects.

This study denoted microbial concentration as fecal contamination which includes coliform, fecal coliform, and E. coli. The microbial concentration unit was measured in either CFU or myeloproliferative neoplasms (MPN). Furthermore, the physical and chemical aspects also followed the standard used in the WHO guidelines of drinking water (WHO, 2017).

Not all data in the three datasets could be analyzed because of some missing data when the household and individual-level surveys were matched. Data included in the IDHS 2007 analysis were 29,840 households (73.3% of the original number of respondents); IDHS 2012 were 31,173 households (71.1%); and IDHS 2017 were 31,253 households (65.2%). Table 2 presents the descriptive statistics for all independent and dependent variables used in binary logistic regression analysis. More than half of the households in IDHS 2007, 2012, and 2017 used non-PDW as a source of daily drinking water. However, the PDW consumer percentage shows an increasing trend from the IDHS 2007 (9.7%) to IDHS 2012 (29.3%) and IDHS 2017 (36.3%).

The highest percentage of the educational levels of parents from three datasets were at the secondary education level. Based on the place of residence, more than half of the respondents lived in rural areas in IDHS 2007 data, whereas more than half of the respondents lived in urban areas in IDHS 2012 and 2017. The use of mass media respondents from IDHS 2017 had the highest average score of 4.25±2.84, IDHS 2007 was 4.10±2.14, and IDHS 2012 was 3.06±1.41 (range 0–9). The majority of people in Indonesia used private toilet facilities for sanitation.

The trends of PDW in each province in Indonesia according to the three datasets are shown in Figure 2. DKI Jakarta and Riau Islands are the provinces in Indonesia that have the highest percentage of PDW consumers. East Nusa Tenggara province consistently has the lowest PDW consumer in three IDHS datasets. Generally, there is an increased PDW usage across all provinces in Indonesia from 2007 to 2017. However, some provinces, such as Aceh, West Papua, Jambi, Bengkulu, and East Nusa Tenggara, have declined, especially from 2012 to 2017.

Logistic regression analysis results from IDHS 2007, 2012, and 2017 were presented in Table 3. The regressions explained 33.3, 28.4, and 29.6% of variances in IDHS 2007, 2012, and 2017, respectively. The age of the household head, mother's educational level, father's educational level, wealth index, type of residence, regions, and type of toilet facility were significantly associated with the PDW usage.

The type of residence, wealth index, and regions were considered to have the largest influence on PDW consumption, i.e., highest β values, in all three dataset comparisons. Based on IDHS 2007, respondents who lived in urban areas were 2.3 times more likely to utilize PDW compared to rural areas, and the value increased to 2.8 in IDHS 2012 and 2.6 in IDHS 2017. The results also indicate that wealthier households tend to consume PDW. A variation of PDW usage was found between regions in Indonesia, in which other regions (Figure 3) were more likely to consume PDW compared to Eastern Indonesia.

Furthermore, the age of the household head was significant to PDW usage. The mother's educational level had a positive relationship with the PDW consumption, as well as the father's educational level. The type of toilet facility was significantly associated with PDW consumption. The variable of mass media exposure had no significant effect on the PDW usage in the three datasets. Moreover, a significant association was found between PDW and diarrhea cases in children using DHS datasets in 2007 (X² [1]=31.901, p≤0.05), 2012 (X² [1]=22.184, p≤0.05), and 2017 (X² [1]=3.985, p≤0.05), in which PDW is associated with lower diarrhea cases in children.

If we combined PDW samples from all studies (n=425), approximately 48% of the total samples (n=214) were fecally contaminated and 25% (n=36) exceeded the pH threshold. These nine studies used either coliform, fecal coliform, or E. coli as microbiological water quality indicators. All studies found fecally contaminated PDW in their samples and samples in two studies were found to exceed the pH standard, i.e., outside the range of acceptable pH values of 6.5–8.5.

Cronin et al. (2017) included most samples (n=183), in which 50.8% were fecally contaminated. Irda Sari et al. (2018) was the second study with most samples (n=126) and revealed that 50% of their refilled water samples were fecally contaminated. Baharuddin & Ichsan (2020) revealed that 100% of their PDW samples (n=6) were fecally contaminated and also exceeded the standard pH. Moreover, Baharuddin et al. (2018) revealed that 80% of their PDW samples (n=30) were fecally contaminated and approximately 10% exceeded the standard pH.

Our results found that PDW consumption in Indonesia continues to increase and is predicted to reach 50% of consumers by 2026. According to SUSENAS 2020 data, refilled drinking water was the most widely used source of drinking water in Indonesia (29.1%). Meanwhile, 10.23% of households used un-refilled water as the main source of drinking water. The increasing use of PDW in Indonesia follows the global situation, e.g., in China, the USA, or European countries. The sales of un-refilled or bottled water reached over 200 billion liters globally in 2007. Europe and North America were the biggest markets, but the sales are expanding in many developing countries (Gleick & Cooley, 2009; Qian, 2018). Contrastingly, a slowing consumption trend of PDW consumers was found in Indonesia from 2012 to 2017 compared to 2007–2012. This may cause a decreased average rate of population growth in Indonesia from 1.9% per year in 2000–2010 to 1.25% per year in 2010–2020 (Statistics Indonesia, 2021).

The analysis revealed that the trend of PDW consumption in Indonesia is strongly correlated with the economic growth level of the country, i.e., proxied by urbanization and GDP levels. We argue that the economic growth of a country indirectly influences one's working time, i.e., increased working time, limits the spare time, and makes them choose a time- and cost-efficient drinking water option, i.e., PDW (Komarulzaman et al., 2017). Additionally, the regression analyses revealed that wealthier households positively correlated with PDW consumption. Households who have higher incomes can afford the PDW cost (Irianti et al., 2016; Adil et al., 2021). We argue that the PDW cost is relatively affordable for the majority of the Indonesian population. The average expenditure on drinking water is IDR 119.850 (USD 8.3) with the assumption that the total water demand per month is 115 L per capita for four people in a household. Based on income, water expenditure for low-income households is 77.556 (USD 5.34), middle-income is 134.375 (USD 9.25), and high-income is IDR 182.391 (USD 12.56) (Prayoga et al., 2021).

Urban households were more likely to consume PDW compared to rural households. Firstly, because the urbanization level increased the accessibility of PDW producers or markets while accessing PDW in rural areas is more difficult due to limited infrastructure (Irianti et al., 2016). Moreover, there is a tendency of the urban population to consume ready-to-use drinking water, i.e., PDW (Baharuddin & Ichsan, 2020; Saimin et al., 2020).

Our study revealed that the educational levels of parents were associated with the likelihood of choosing PDW as the main drinking water source. Similar findings were found in other studies from Ethiopia, Pakistan, and Indonesia (Nastiti et al., 2017; Adil et al., 2021; Gebremichael et al., 2021). The increased educational levels of parents may lead to increased beneficial awareness of PDW as the main source of drinking water, e.g., quality, convenience, and affordability.

Region or location is significantly related to PDW consumption. In the IDHS 2007 and 2012, respondents who lived in Java and Bali relatively consume PDW more than other regions, probably because Java and Bali are more developed and urbanized than other regions in Indonesia. However, there was a remarkable increase in PDW consumption in the Borneo region, from 7% in IDHS 2007 to 47.2% in IDHS 2017, possibly because of a quite significantly increased economic growth in this region (Afkarina et al., 2019). Meanwhile, the peat swamp areas are commonly found on the Borneo island, which are characterized by a low pH, high turbidity, and high organic content (Oktiani et al., 2020). People on this island may see that PDW is the best option for their drinking water source. PDW consumption in some provinces declined, e.g., Aceh, West Papua, etc., especially in 2012–2017. We assumed that the fall was caused by the fluctuating GDP in those provinces from 2012 to 2017. Furthermore, the global economic crisis in 2015 may affect the GDP in some provinces in Indonesia, thereby decreasing the PDW consumption. However, further investigation is necessary to ensure the reason for the decreased PDW consumption in these provinces.

Exposure to mass media does not significantly relate to PDW consumption. Particularly, advertising PDW in newspapers, television, and radio may be of no influence or have a small correlation to choosing PDW. The findings support the study of Doria (2010) that impersonal information, e.g., mass media, does not have a direct influence on PDW consumption compared to social-economic status and interpersonal information from family or friends. The increased PDW consumption in the neighborhood may create a norm, i.e., social pressure, of using PDW, which will create a ‘reinforcing effect’ and rapidly increase the PDW consumption, as discussed in the context of household water treatment (HWT) behavior in developing countries (Daniel et al., 2022).

Another significant variable is the household head's age. A negative coefficient indicates that younger household head tends to consume PDW, probably because the young household head does not want to bother themselves with the time to do HWT, so they can focus more on their job. Income, educational level, and urban area positively influence PDW consumption in combination with this finding with other study findings, we then argue that young people in urban areas may dominate the PDW consumer in the future. These young people can then be a potential object of intervention in improving PDW safety in Indonesia, e.g., by educating them to always keep their PDW dispensers clean.

The SLR revealed that approximately half of the analyzed samples were fecally contaminated. However, the Ministry of Health of Indonesia (Ministry of Health, 2021) revealed that 67% of the refill water was contaminated. Some studies imply that PDW has relatively better water quality than other types of drinking water sources, e.g., tap, protected well, spring, etc., although PDW is not fully safe (Cronin et al., 2017; Irda Sari et al., 2018). First, the treatment process is not effective. A previous study found that a combination of water treatment in refilled PDW depot, e.g., reverse osmosis, ultraviolet, and ozone, decreases the total coliform by 92.6%, suggesting that the treatment process is not fully effective. Secondly, there is a high possibility of recontamination after treatment, e.g., during packaging or consumption at the house (Sari et al., 2020). Other factors that influence the water quality are hygienic processes during production, improper storage, high temperature, lack of protection after treatment, and the quality of raw water sources, which may be caused by open defecation, underground damaged sewerage lines, and drainage system seepage (Halage et al., 2015).

The systematic review revealed that PDW is not always safe although people often perceive that PDW has a good quality. Therefore, boiling water treatment is recommended to ensure the safety of PDW. Waterborne pathogen exposure will be significantly reduced and be safer if a household boils its drinking water (Cohen et al., 2020). Another strategy is to make sure that the water dispenser, the area surrounding the dispenser, and the water cup are hygienic, e.g., by not putting the bottle or dispenser on the floor, as shown by the rightmost picture in Figure 6.

Furthermore, issues of microplastic contamination in PDW are rising in recent years. Microplastics have been identified in bottled water. The presence of microplastics in PDW has been highly detected in reusable plastic bottles (Eerkes-Medrano et al., 2019). The contamination of microplastic in PDW could also occur in Indonesia due to increased PDW consumption. A study (including Indonesia) reported that 93% of samples (n=259) of bottled water from across 11 different brands showed some sign of microplastics. The average density of microplastics from all samples is 325 MPP/L for (size: >100 μm) and 315 MPP/L (size: 6.5–100 μm) (Mason et al., 2018).

The safety aspect of PDW cannot be underestimated. Unsafe drinking water threatens human health and can lead to morbidity and in some cases, death. Diarrhea was the most common disease that occurred through water transmission, especially among children, who are the most vulnerable group. Approximately 11% of child deaths in the world are linked to diarrhea from unsafe drinking water (Pal et al., 2018). Our statistical analyses confirm that PDW is associated with lower diarrhea cases in children although PDW is not always safe, which could be because other types of water sources have worse water quality than PDW and are not treated. However, improving the quality of PDW and minimizing recontamination to reduce the risk of getting diarrhea in children is important because children who consume PDW suffer from diarrhea.

HWT, especially boiling, is practiced by the majority of Indonesian households. However, HWT consumption has decreased in the past decade. The IDHS reports show a decreased HWT, i.e., 91% in 2007 compared to 70% in 2017 (National Population and Family Planning Board et al., 2018). Boiling is considered a time-consuming HWT (Clasen et al., 2008). The decreasing consumption of HWT is due to the increasing PDW consumption. However, this needs more investigation.

This study has some limitations. We realize that other variables may affect the household decision to consume PDW but are not included in this study, especially perceptions regarding PDW (Gebremichael et al., 2021). Perceptions or psychological factors are considered to have higher explanatory power to explain water-related behavior compared to socio-economic characteristics (Lilje & Mosler, 2017). Future studies should investigate the household's perceptions regarding PDW to provide a better understanding of PDW consumption in Indonesia. Furthermore, future studies need to confirm our argument that the decreased HWT is due to the increasing PDW consumption. We could not confirm this argument because of the limited annual data on HWT in Indonesia, unlike PDW. One can use data from other countries to confirm this argument. Moreover, we need more studies to find the reason for many contaminated PDWs. The water quality analysis combined with the sanitary inspection at the water producer, depot, and house, is expected to reveal this issue in more detail. We need to investigate where (e.g., depot or house) and when (e.g., during treatment, production, or consumption at the house) contamination or recontamination occurs. Future studies also need to discover the role of mass media or other communication channels to influence the household decision in consuming PDW. This can help us to understand how people change their main drinking water source to PDW.

Finally, this study is limited due to the use of secondary data. We have relied on other studies or data sources and have no control over the data collection or reporting. Particularly, there is an anomaly in the data of PDW consumers in Indonesia in 2012, i.e., the percentage significantly increased from 22.29% in 2011 to 38.85% in 2012, but then decreased to 27.66% in 2013 (Figure 3). This raises a question in either data collection, validation, or reporting by the related agencies. Another example is PDW water sampling. Information on sample collection and testing is limited in previous studies, either directly from the sealed bottle, from the water dispenser (directly to the sampling bag), or the cup. This may moderate (or confirm) our conclusion that one out of two PDW in Indonesia is fecally contaminated.

This study reveals the trends of PDW consumption in Indonesia in the past decades. There is a fast-increasing PDW consumption in Indonesia and 50% of people in Indonesia are expected to consume PDW in 2026. The increasing PDW consumption in Indonesia was strongly associated with the economic growth of the country, which is represented by the GDP and urban population. Regression analysis revealed that socio-economic characteristics, including the age of the household head, mother's educational level, father's educational level, wealth index, type of residence, and type of toilet facility, significantly predict the PDW consumption. Our findings indicate that young people in urban areas may dominate the PDW consumer in the future. Moreover, past studies revealed a high chance of fecal contamination in the PDW, suggesting the need to better regulate and implement the hygienic procedure of PDW production, distribution, and storage, i.e., before reaching the consumer, in Indonesia. Finally, people can perform HWT and make sure that the water dispenser and the surrounding area are hygienic to prevent recontamination at the household level.

The first author receives a master's study funding from Indonesia Endowment Fund for Education (LPDP). The second author receives master's study funding from the Balikpapan Stimulan Scholarship.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.