Sustained use of liquefied petroleum gas following one year of free fuel and behavioral support in Puno, Peru

Abstract

Background:

Existing efforts to promote cleaner fuels have not achieved exclusive use. We investigated whether receiving 12 months of free liquefied petroleum gas (LPG) and behavioral support could motivate continued purchase and use.

Methods:

The Cardiopulmonary outcomes and Household Air Pollution (CHAP) trial enrolled 180 women. Half were randomly assigned to an intervention group, which included free LPG delivered in months 1–12 followed by a post-intervention period in which they no longer received free fuel (months 13–24). For the purposes of comparison, we also include months 1–12 of data from control participants. We tracked stove use with temperature monitors, surveys, and observations, and conducted in-depth interviews with 19 participants from the intervention group at the end of their post-intervention period.

Results:

Participants from the intervention group used their LPG stove for 85.4 % of monitored days and 63.2 % of cooking minutes during the post-intervention months (13–24) when they were not receiving free fuel from the trial. They used a traditional stove (fogón) on 45.1 % of days post-intervention, which is significantly lower than fogón use by control participants during the intervention period (72.2 % of days). In months 13–24 post-intervention, participants from the intervention group purchased on average 12.3 kg and spent 34.1 soles (10.3 USD) per month on LPG. Continued LPG use was higher among participants who said they could afford two tanks of LPG per month, did not cook for animals, and removed their traditional stove. Women described that becoming accustomed to LPG, support and training from the project, consistent LPG supply, choice between LPG providers, and access to delivery services facilitated sustained LPG use. However, high cost was a major barrier to exclusive use.

Conclusion:

A 12-month period of intensive LPG support achieved a high level of sustained LPG use post-intervention, but other strategies are needed to sustain exclusive use.

Introduction

Nearly 2.5 billion people worldwide use primarily unprocessed biomass fuels such as wood, dung, and charcoal for cooking (IEA, 2022). Household air pollution (HAP) produced by burning these biomass fuels directly in the household leads to many adverse health consequences, such as low birth weight, pneumonia, chronic obstructive pulmonary disease, lung cancer, and cardiovascular disease (GBD 2019 Risk Factors Collaborators, 2020). Additionally, household-level use of biomass fuels for cooking has negative environmental, social, economic, and quality of life consequences. Cooking with biomass can reduce the amount of time people have available for economically productive activities or for sleep, cause discomfort due to burning eyes or coughing from the smoke while cooking, and diminish desired cleanliness of hands, clothes, pots, and kitchens (Williams et al., 2020a; Williams et al., 2020b). Household use of biomass fuel also contributes to outdoor air pollution, deforestation, and climate change (Bailis et al., 2015; Holdren et al., 2000; Martin et al., 2014).

Transitioning to cooking with cleaner fuels, such as liquefied petroleum gas (LPG), ethanol, or electricity, can reduce the burden associated with biomass fuel use (Bruce et al., 2015). However, realizing the full potential benefits of cleaner fuels requires that they be used near-exclusively (Johnson & Chiang, 2015). To achieve this, households must be willing to overcome barriers to adapt to a different cooking technology and fuel, and be able to afford the amount of fuel required for near-exclusive use.

To increase affordability of cleaner fuels, several low- and middle-income country (LMIC) governments have implemented programs that provide free or subsidized cleaner energy technologies and fuel to households in need (Quinn et al., 2018). However, these efforts have not achieved exclusive LPG use, largely because households were unable to overcome the social and behavioral barriers to cooking with cleaner fuel and/or could not afford the amount of cleaner fuel required for exclusive use even at the subsidized prices (Shankar et al., 2020).

The Cardiopulmonary outcomes and Household Air Pollution (CHAP) trial, an individually-randomized controlled trial in Puno, Peru in which participants assigned to the intervention group received free unlimited deliveries of LPG for 12 months (Fandino-Del-Rio et al., 2017), demonstrated that when both monetary and time costs of LPG use were removed, participants used LPG near-exclusively (Checkley et al., 2021; Williams et al., 2020a). However, the intensive support provided by the trial, which included around 20 kg and two home-based deliveries of LPG per month for one year (Williams et al., 2020c), is not sustainable or feasible to replicate at scale. Based on the diffusion of innovation theory, it is possible that a short trialability period of intensive support for LPG use, such as that provided by the CHAP trial, could motivate households to continue purchasing and using LPG in a sustained manner after the support ends (Rogers, 2003). However, to date, no studies have investigated the extent to which participants continue to use LPG after a limited period of receiving free fuel, or the factors influencing such sustained use.

We sought to investigate the impact of providing a free LPG stove and one year of free, unlimited LPG fuel and deliveries on longer-term sustained LPG use by continuing to follow participants from the CHAP trial intervention group for 12 months post-intervention. In this paper, we explore the amount of cooking done with LPG and barriers and facilitators to sustained LPG use over a 12-month period after participants stopped receiving free LPG from the trial, with the goal of informing the strategic design of policies and programs that can promote long-term LPG use for achieving health, environmental, and other benefits.

Methods

Study setting

This study was conducted in eight rural communities of Puno, Peru. Puno is located in southern Peru bordering Lake Titicaca at an altitude of 3825 m. Most rural Puno residents are indigenous Aymara or Quechua and earn a living through farming and raising livestock. Cooking is traditionally done with wood or dung in a three-burner clay stove called a fogón. Recent governmental programs providing free LPG stoves (Cocinas Peru) and subsidized LPG (Fondo de Inclusión Social Energético, FISE; Social Energy Inclusion Fund in English) to poor households have increased access to cleaner energy in the area (Pollard et al., 2018). However, most households continue to use a combination of LPG and biomass fuel for cooking because the amount of subsidized fuel is insufficient to meet all cooking needs (Pollard et al., 2018; Williams et al., 2020c).

Study design

The CHAP trial aimed to assess cardiopulmonary health outcomes, household air pollution exposure, and nutrition outcomes associated with an LPG intervention, which included a free stove, free fuel, and behavioral support over 12 months (Fandino-Del-Rio et al., 2017). We enrolled 180 adult, non-pregnant women between 25 and 64 years of age who cooked daily with biomass fuels. The study enrolled only women given formative research findings that most cooking is done by women in the study area (Hollada et al., 2017), as is similarly true in other LMICs (Bonjour et al., 2013). We individually randomized participants to intervention or control groups (n = 90 per arm) on a rolling basis from March 1, 2017 to February 15, 2018 (see Fig. 1). After randomization, intervention group participants received a new 3-burner LPG stove and training on how to use it safely, a 10 kg LPG tank, a cooking demonstration, education on the benefits of LPG, and unlimited free LPG refills delivered to their homes for 12 months. Control group participants were encouraged to continue their baseline cooking practices during months 1–12 after randomization. For months 13–24 after randomization, participants from the intervention group stopped receiving free LPG deliveries yet kept the trial-provided LPG stove and tank. During months 13–24 after randomization, control group participants received nearly the same intervention package as was provided to the intervention group during the trial’s first 12 months. However, instead of free LPG delivered to their homes, control group participants received vouchers for the full cost of two LPG tanks (20 kg) per month. Two tanks were deemed sufficient for exclusive LPG use based on average LPG consumption rates by intervention group participants in months 1–12 of the trial (Williams et al., 2020c). Given the staggered enrollment, months 13–24 of follow-up began on March 1, 2018 for participants enrolled first and ended on February 15, 2020 for those enrolled last.

Fig. 1.

Schematic depicting the CHAP trial design.

Stove use monitoring

We monitored both biomass and LPG stoves in homes of participants from the intervention group for the entire duration of the 24-month trial using LabJack Digit-TL stove use monitors (SUMs), unless biomass stoves were removed (LabJack, n.d.). We also installed SUMs on biomass stoves of all control group participants during months 1–12. Fieldworkers visited participant homes every month to download the temperature data and check the devices. SUMs were installed in newly constructed fogones and removed from dismantled fogones. At each SUMs download visit in months 13–24, fieldworkers also administered a survey on the number of LPG tanks purchased, use of governmental subsidies for LPG, the amount of money spent on LPG, and repairs made to the LPG stove in the previous month.

Quantitative data collection

For all participants, fieldworkers conducted a survey at baseline, and 12- and 24-months post-randomization on socio-demographic characteristics such as number of household members, animals owned, education, asset ownership, enrollment in FISE, and ownership of LPG stoves and tanks acquired by participants from outside the trial. The 24-month survey additionally included questions on the condition of the trial-provided LPG stove, perceptions of the price of LPG, time spent obtaining LPG, and cost of LPG delivery.

During visits to all participants at 15, 18, 21, and 24-months post-randomization, fieldworkers also administered a questionnaire detailing the number and length of cooking periods and types of stoves used. This survey included questions about reasons for liking and disliking the LPG stove, time spent collecting biomass fuel, problems with the LPG stove, and maintenance performed on the LPG stove. All questionnaires were administered on tablets using the Research Electronic Data Capture (REDCap) data collection and management software (Harris et al., 2009).

Qualitative data collection

One fieldworker trained in qualitative methods conducted 19 in-depth interviews with participants from the intervention group near the end of their post-intervention period (Table 1). Using SUMs data, we selected participants with varying rates (lower/higher) of continued post-intervention LPG use and sought to include some FISE beneficiaries and some non-beneficiaries. Interviews covered themes including opinions about LPG, cooking practices, intervention impact on cooking behaviors, comparison of cooking behaviors during and after receiving free LPG, process for obtaining LPG refills, participation in and perceptions of FISE, and factors influencing stove use decisions. We explored reasons for abandonment or continued use of LPG once it was no longer provided for free.

Table 1.

Characteristics of in-depth interview participants from the intervention group.

Participant	% cooking with LPG	Wealth quintile	FISE	Household size	Number of pigs/dogs	Fogón in house	Owns 2 LPG tanks
1	51.5 %	2	Yes	3	1	Yes	Yes
2	100 %	2	Yes	4	2	No	Yes
3	56.5 %	2	Yes	8	3	Yes	Yes
4	93.8 %	1	Yes	4	1	No	Yes
5	100 %	1	Yes	3	1	Yes	Yes
6	99.1 %	2	Yes	4	7	Yes	Yes
7	73.1 %	2	Yes	2	2	Yes	Yes
8	82.7 %	3	Yes	5	0	No	Yes
9	50.5 %	2	Yes	2	1	Yes	Yes
10	53.6 %	1	No	2	4	Yes	Yes
11	32.1 %	2	No	6	2	Yes	Yes
12	100 %	1	Yes	3	1	Yes	No
13	58.0 %	1	Yes	3	2	Yes	Yes
14	69.2 %	2	Yes	4	1	Yes	Yes
15	74.4 %	2	No	3	1	Yes	Yes
16	53.1 %	1	Yes	6	4	Yes	Yes
17	31.3 %	2	Yes	5	2	Yes	No
18	95.7 %	1	No	4	1	Yes	Yes
19	75.5 %	2	No	3	2	Yes	No

Data analysis

Our analysis focuses on the data collected from intervention group participants during the post-intervention period (months 13–24 post-randomization). However, we also include data on stove use by intervention and control participants during the intervention period (months 1–12) for comparison purposes.

Quantitative data were analyzed using StataSE version 15 (Stata-Corp, 2017) and R version 3.6.2 (R Core Team, 2022). We analyzed the SUMs data using a previously published algorithm to identify cooking events and minutes spent cooking based on temperature increases and decreases (Williams et al., 2020b). We slightly adjusted the previously published algorithm to identify SUMs errors (strings of constant 0 degree Celsius temperature readings), and dropped data for days on which errors were flagged. For four participants who removed their fogón during the post-intervention period as verified through surveys, photos, and consistent LPG use post-removal with no gaps, we set fogón use to zero for all days after the removal date. For 7 participants who removed their fogón during the intervention period, we set fogón use to zero after the point of removal until LPG use dropped to zero; we assumed fogón SUMs data to be missing after that point. Adjustments made to SUMs data from the intervention period (months 1–12) have been previously explained (Williams et al., 2020b).

We calculated two separate LPG use metrics. First, we analyzed the LPG stove data individually, calculating the percent of monitored days in which the LPG stove was used. Second, using only data from days in which both the fogón and LPG stove were monitored, we calculated the percentage of total cooking minutes done with LPG. We additionally used the data from SUMs on fogones to calculate the percent of monitored days with fogón use. We used t-tests to compare between intervention and control groups. After visual inspection of the plot of LPG use by month, we observed that LPG use seemed to decline rapidly after the end of the intervention then level off. Thus, we conducted a series of linear regressions of month on percent of cooking done with LPG, first including all data (months 12–24), then eliminating month 12 data, and so on. We considered the decline to stabilize when the month variable was no longer statistically significantly related to percent of cooking done with LPG.

We also used linear regression to investigate factors influencing the average percent of cooking done with LPG post-intervention by participants from the intervention group (continuous outcome variable). The model included continuous variables (years of education and number of household members), dichotomous variables (reported ability to afford 2 LPG tanks per month, ownership of pigs or dogs who typically eat cooked food (Williams et al., 2020c), participation in FISE, maintaining a fogón in the house, and ownership of two LPG tanks), and a categorical variable for wealth quintile (lowest, low, and middle). Differences in the percent of cooking done with LPG based on the value of each of these variables are shown in Table 5. Stata output from the model is included in Appendix B.

Table 5.

Quantitative factors affecting percent of cooking done with LPG in the post-intervention period among participants from the intervention group (n = 80).

	Change in percent of cooking done with LPG in post-intervention period (by levels of variables in the table)	p-Value
Owns pigs or dogs	− 26.3	<0.001
Reports ability to afford 2	18.8	0.004
LPG tanks per month
Fogón still in the house	− 20.2	0.009
Wealth quintile
Lowest (ref) (n = 29)
Low (n = 49)	2.5	0.63
Middle (n = 2)	21.6	0.18
Has 2 LPG tanks	− 5.2	0.36
FISE participant	− 2.1	0.70
# household members	− 0.7	0.62
Years of education	0.04	0.95

We calculated descriptive statistics to summarize LPG purchasing behaviors and perceptions, and conducted Chi-square tests to analyze differences between groups. We also used survey data on reported stove use to calculate the percent of reported cooking minutes done with LPG, and compared between groups using t-tests.

The first author reviewed all qualitative interview transcripts to identify key emerging themes related to sustained LPG use. Themes were discussed and agreed upon by a group of co-authors. The first author coded the transcripts according to the identified themes and reviewed coded quotes with the co-author group. Any differences in opinion regarding coding assignment were resolved through discussion. Coded quotes were then examined and compared to understand how each theme influenced sustained LPG use, and the relationships between themes. We used Taguette, an open-source qualitative data analysis program, to assist with data management, indexing, and coding (Rampin & Rampin, 2021).

Results

Participant characteristics

On average, participants from both intervention and control groups had about 6 years of education and 4 household members at the beginning of the post-intervention period (12 months after randomization) (Table 2). The majority of participants were in the two lowest wealth quintiles. Around half of participants owned pigs and nearly three-quarters owned dogs, animals for which people typically prepare cooked food (Williams et al., 2020c). Just under half of intervention group participants and just over half of control group participants were currently or previously enrolled in FISE. Approximately 70 % of intervention group participants had two LPG tanks in their household, one provided by the trial and the second received or purchased from outside the trial.

Table 2.

Characteristics of intervention and control group participants at the conclusion of the intervention period (12 months post-randomization).

	Intervention		Control
	n	% or mean (SD)	n	% or mean (SD)
Years of education	89	6.1 (3.4)	90	6.4 (3.3)
Wealth quintile
1 (lowest)	33	37.1 %	37	41 %
2 (low)	51	57.3 %	50	55.6 %
3 (middle)	5	5.6 %	2	2.2 %
4 (high)	0	0 %	1	1.1 %
Number of household members	89	3.6 (1.7)	90	3.5 (1.5)
Number of pigs owned
0	48	53.9 %	47	52.2 %
1–2	36	40.4 %	38	42.2 %
3+	5	5.6 %	5	5.6 %
Number of dogs owned
0	22	24.7 %	29	32.2 %
1	43	48.3 %	34	37.8 %
2+	24	27.0 %	27	30.0 %
Household was ever enrolled in FISE	36	40.5 %	48	53.3 %
Owns two LPG tanks	62	69.7 %		N/A

Sustained LPG use

During the intervention period (months 1–12), participants in the intervention group (who were receiving free LPG) used the LPG stove on 98.7 % of days and used LPG exclusively on 96.7 % of days. They used LPG for 98.2 % of cooking minutes and 99.2 % of cooking events (Table 3).

Table 3.

Percent of days, minutes, and events cooked by stove type by intervention group participants during the intervention (when they were receiving free, unlimited LPG) and post-intervention (when they no longer received any LPG from the trial).

	Intervention (months 1–12)		Post-intervention (months 13–24)
	n	% or mean (SD)	n	% or mean (SD)
Using only LPG SUMs data
Average # of days with LPG stove monitored, mean (SD)	90	357.5 (18.0)	89	345.5 (45.7)
% days with one or more LPG event	90	98.7 %	89	85.4 %
Average # LPG events per day, mean (SD)	90	2.7 (0.4)	89	2.1 (0.7)
Average minutes spent cooking with LPG per day, mean (SD)	90	183.2 (90)	89	130.2 (47.9)
Using only data from days with both LPG and fogón monitoring
Average # of days with both LPG and fogón stoves monitored, mean (SD)	90	353.7 (19.2)	83	315.3 (66.0)
% days with exclusive LPG use	90	96.7 %	83	51.1 %
% days with stove stacking (both LPG and fogón used)	90	2.0 %	83	35.6 %
% days with exclusive fogón use	90	0.1 %	83	9.7 %
% days with no cooking	90	1.3 %	83	3.6 %
% of cooking minutes done with LPG	90	98.2 %	83	63.2 %
% of cooking events done with LPG	90	99.2 %	83	76.8 %

In the post-intervention period (months 13–24), participants from the intervention group used the LPG stove on 85.4 % of monitored days for an average of 2.1 events per day (Table 3). They used LPG exclusively on around 51 % of days and in combination with the fogón on around 36 % of days (with exclusive fogón use on 9.7 % and no cooking on 3.6 % of days).

Participants from the intervention group who had both stoves monitored during the post-intervention period (n = 83) used LPG for 63.2 % and 64.9 % of cooking minutes post-intervention according to SUMs and self-report, respectively (Tables 3 and 4). This amount of LPG use is lower than the near-exclusive use (98.2 %) recorded by SUMs among the same participants during the intervention period (months 1–12) when they were receiving free LPG (p < 0.001; see Fig. 2). However, it is significantly higher than the 20.2 % of cooking minutes reportedly done with LPG by control group participants who owned LPG stoves but were not receiving free fuel in months 1–12 of the trial (n = 80; p < 0.001; Table 4). Additionally, data from SUMs on fogones showed that control group participants who owned an LPG stove from outside the trial (n = 80) used the fogón on 72.2 % of days during months 1–12 (when they were not receiving free LPG), while participants from the intervention group used the fogón on only 45.1 % of days post-intervention (when they were not receiving free LPG) (p < 0.001).

Table 4.

Self-reported cooking with LPG and SUMs-recorded fogón use, among intervention group participants during the intervention (when they were receiving free LPG from the trial), participants from the intervention group after the intervention (when they were not receiving free LPG from the trial), and control group participants (who were not receiving free LPG from the trial).

	Intervention (months 1–12)		Post-intervention (months 13–24)		Control (months 1–12)
	n	%	n	%	n	%
Using survey data
% of reported cooking minutes done with LPG	90	99.3 %	89	64.9 %	80	20.2 %
% of reported cooking events done with LPG	90	99.4 %	89	69.7 %	80	28.0 %
% of participants reporting LPG as their primary stove (at the end of each period, 12 or 24 months after randomization)	89	98.9 %	86	67.4 %	80	8.8 %
Using only fogón SUMs data
Average # of days with fogón monitored, mean (SD)	90	358.4 (18.5)	83	322.7 (64.9)	90	347.1 (20.3)
% days with no fogón use	90	97.9 %	83	54.9 %	90	27.8 %
% days with one or more fogón event	90	2.1 %	83	45.1 %	90	72.2 %
% days with two or more fogón events	90	0.2 %	83	15.5 %	90	24.7 %
Average # fogón events per day, mean (SD)	90	0.02 (0.04)	83	0.6 (0.5)	90	1.0 (0.4)
Average minutes spent cooking with fogón per day, mean (SD)	90	3.6 (6.1)	83	91.9 (74.6)	90	153.0 (68.8)

Fig. 2.

Percent of cooking done with LPG among participants from the intervention group during the intervention (months 1–12) and post-intervention (months 13–24).

Notably, LPG stove use by participants from the intervention group declined rapidly in the first two months after the intervention ended (Fig. 1); the decline stabilized at month 15 (slope not significantly negative, p = 0.09). In months 15 through 24, intervention group participants used LPG for an average of 61.3 % of cooking minutes. Additionally, there was great variability in the percent of cooking done with LPG across participants and month-to-month (Fig. 2).

Compared to the percent of cooking minutes, the percent of cooking events done with LPG by participants from the intervention group post-intervention (months 13–24) was even greater according to both SUMs (76.8 %) and survey data (69.7 %) (Table 3). This is likely because each event on the traditional stove (158 min on average) is longer than each event on the LPG stove (68 min on average).

All participants from the intervention group purchased LPG fuel at some point over the post-intervention period (months 13–24). On average, they purchased 1.2 tanks (12.3 kg) of LPG per month (SD 0.4; range 0.4–2.3). Across all participants from the intervention group, 51.1 % (n = 45 out of 88) bought at least one tank of LPG every month during the post-intervention period. Overall, participants from the intervention group purchased LPG in 88.6 % of post-intervention months (SD 16.1 %; range 41.7 %–100 %) (10.6 months out of 12).

Participants from the intervention group spent on average 34.1 soles (10.3 USD) per month on LPG during the post-intervention period (SD 13.3; range 6.1–71.2). Considering that the cost of an LPG tank after applying the FISE discount was 20 soles (6.1 USD), 96.6 % of participants paid more than the discounted price of LPG in at least one month of the post-intervention year. On average, participants from the intervention group paid more than the discounted LPG tank price in 7.6 months of the post-intervention year.

Factors influencing sustained LPG use

Participant characteristics and perceptions

The percent of cooking done with LPG by participants from the intervention group in the post-intervention period was significantly associated with reported ability to afford two LPG tanks per month, cooking for animals, and maintaining a fogón in the home (Table 5). Measured as a percentage of cooking minutes, LPG use was 19 % higher among participants who reported that they could afford two tanks of LPG per month, but 26 % lower among those who cooked for animals and 20 % lower for those who maintained a fogón in their house. Years of education, participation in the FISE program, number of household members, and number of LPG tanks owned (1 or 2) had no statistically significant influence. However, fewer than 30 % of participants had only one LPG tank in their household. We also found no statistically significant relationship between wealth and sustained LPG use, although this could be because most participants were in the two lowest wealth quintiles.

Experience with LPG

Over the course of the intervention period, participants observed the benefits and became accustomed to cooking with LPG. In in-depth interviews, participants described how they saved time because LPG cooks more quickly, they could do other things while the food was cooking, and that LPG was easy to use. Participants mentioned that when they cooked with the fogón, they always smelled like smoke and felt ashamed to be around other people; they had to bathe and wash their clothes more often because of the blackening and smokey smell, and thus spent more money on soap. With LPG, they felt cleaner and more confident. Additionally, they no longer experienced burning eyes or coughing from the smoke, and they did not have to collect dung or wood.

Because of these observed benefits, participants described how they became “accustomed to gas” (Participant 2) and “too lazy to cook with the fogón ” (Participant 4). Those who had LPG before the project typically only used it in the afternoons when they arrived home tired after a day in the field, or when it was raining or they were in a hurry. After becoming accustomed to using LPG during the trial, they continued to use LPG primarily, even though they described their financial situation as similar to before the project when they were using primarily biomass. Many participants reported that their experience caused them to prioritize spending on LPG more highly and do whatever they could to find the money to buy it.

“I’m always using gas. It is cleaner, faster and I’m accustomed to it. What can I do? I need to get the money one way or another.”

(Participant 15)

With LPG, most women said they saved time on fuel collection because they could easily get a refill delivered or other household members would exchange the tank. In contrast, other household members typically did not help collect biomass fuel.

“I go to the store myself… If they deliver it, they charge an extra sol. I don’t want to pay that, so I go to the store and I bring it here myself… Sometimes my children help, since they’re young (18, 20, and 21)… But they don’t collect dung. They’re lazy. They say their hands will get dirty so they don’t want to… But when the gas runs out, I tell them to go, and they go.”

(Participant 4)

Even participants who returned to using the fogón as their primary stove did so begrudgingly; they described extreme dislike of cooking with the fogón, but said they had no choice given their financial situation. Many described using strategies such as cooking outside and leaving the pot alone while it was cooking to make using the fogón more bearable.

“I feel bad because I have to cook with the fogón. The project gave me gas for one year and I got accustomed to it. When the project ended… I had to go back to the fogón…. I felt bad, but what can I do? There’s no money and I have to cook.”

(Participant 16)

Many participants said they continued using LPG despite difficulties with exchanging the tanks or problems with the LPG stove or tanks. They described their willingness to overcome the barriers because, from their experience, the benefits outweighed the challenges.

Training and support

The training and ongoing advice that participants received from the project fieldworkers during the trial helped them overcome fears and concerns that previously hindered their adoption of LPG. For example, women described how they were previously afraid of tank explosions, but now knew how to prevent them by using LPG safely, buying high quality tanks, and properly maintaining, repairing, or replacing the stove parts.

“[Before the project], we were afraid, because on the radio they said that it causes fires. But here [in the project] they taught us, they trained us, and we are no longer afraid. I was afraid mostly of turning on the gas stove, but since we have been trained, I am not afraid anymore.”

(Participant 11)

By the end of the trial, 87 % of intervention group participants (n = 75 out of 86) reported in surveys that they had no fears about LPG; 13 % reported being afraid of gas leaks and 1 % of changing the LPG tank. In interviews, women described learning that they needed to periodically replace certain stove parts to avoid problems, which allayed their fears. During the trial, the fieldworkers taught the women how to check for gas leaks and what to do in the event of a leak. This knowledge helped women feel empowered to not only respond to the leak, but also to stand up to providers who resisted replacing the faulty tank. This feeling of empowerment allowed women to recognize and respond to problems, instead of being afraid and discontinuing LPG use.

Despite the education from the project, some women were still afraid to connect the LPG tanks themselves. Women reported that most local LPG distributors did not provide any advice or training, and most deliverers did not install or test their LPG tanks. Almost all women described feeling more confident about using LPG when the project fieldworkers installed the tanks and verified that everything was working properly. Some women reported asking the LPG deliverers to install and check the tank for them, and said they prefer providers who are willing to provide that service. Receiving this verification made women feel more confident and safer to continue cooking with LPG.

“The fieldworkers always came and reviewed everything. They weighed the tank and installed it. Now, I have to install it. The distributors bring it to my house and leave it outside. Before, the fieldworkers checked everything. Now, I’m afraid.”

(Participant 17)

Choice between many LPG distributors

All interviewees mentioned the large number of LPG distributors available and competing for customers in the region, which allowed them to choose a high-quality and responsive distributor.

“The providers also call us, saying ‘we can bring you gas.’ They give us their numbers. The providers around here fight to get clients from other providers. There are several providers here.”

(Participant 6)

Women preferred distributors who offered flexible payment options, such as providing LPG in advance of payment or before returning an empty tank. Some also allowed a family member to exchange a FISE voucher normally only usable by the person named on it. Other distributors allowed customers to select their own tank from existing stock; women appreciated this because they were able to choose a tank they perceived as heaviest and thus most full. Many women, who said they are usually in charge of exchanging LPG tanks, reported preferring female distributors whom they could befriend and trust.

“Sometimes I ask the lady of the store for a favor. I say, ‘My voucher hasn’t arrived, but I ran out of gas. Can you please give me a tank in advance, I will bring you the voucher later.’ Luckily, she agrees to do this. So she gives us the tank and later when I receive the voucher, I give it to her.”

(Participant 5)

Since almost all distributors accepted FISE vouchers, women could easily switch if they were dissatisfied with service. Participants noted that many distributors offered small household goods or products as gifts to encourage continued patronage.

Delivery availability

Women often identified the time and effort to obtain LPG refills as a barrier to LPG use. In surveys, participants from the intervention group reported requiring an average of 60 min including travel time (SD 36.8, range 4–180) to exchange their tank during the post-intervention period. Because of this, delivery availability was a key facilitator of continued cooking with LPG. In interviews, several women reported that deliveries saved them time and transport costs, and freed them from having to carry a heavy tank. Some distributors provided free delivery to regular customers who lived nearby. However, many women said they were willing to pay extra for the convenience of delivery. Additionally, the fact that deliveries were usually easily available and quick meant women could wait until their tank was completely empty before calling for a refill, thus minimizing waste and unnecessary spending.

“I don’t want to go to Acora [nearby town where distributors are located]. It takes time, I lose time. And it’s heavy to carry the tank. So I just call. They charge a fee, but they bring it.”

(Participant 17)

However, deliveries were sometimes delayed or unavailable, in which case women either resorted to using their fogón, carried the heavy tank to a store to exchange it themselves, or waited for their husband or some other family member to exchange it. Deliveries were more often delayed for people who lived farther away and in the rainy season when flooding could make it difficult to reach rural houses. Additionally, most distributors would not deliver after dark. Also, obtaining a delivery required access to a cell phone and credit to call, which women sometimes did not have.

“He says, ‘I’ll bring it first thing in the morning.’ But they always delay in bringing it. He says he’ll bring it at 6am, but he comes at 7:30am. And since it’s late, we have to cook with the fogón.”

(Participant 1)

LPG stove durability and supply stability

LPG stove durability was high throughout the 24 months after initial delivery, indicating that sustained LPG use was not impacted by stove failure. In the survey conducted 24 months after the LPG stove was delivered, 94.2 % (n = 82) of intervention participants’ stoves had 3 working burners; 5.8 % (n = 5) only had two working burners (in all cases, the middle burner was not working). After 24 months, 100 % of stoves had at least two working burners.

Women also stated that LPG was always available at a standard price. Shortages were rare, and participants could easily go to a different distributor if one ran out of LPG. Because Peru’s LPG system allows any brand or style of tank to be exchanged for any other brand or style, women did not have to worry about whether a distributor would accept their tank. No participant reported any brand preference, stating that they accepted whatever tanks the distributors carried.

Ownership of two LPG tanks

Women reported that owning two LPG tanks facilitated more consistent use. Switching to a second tank as soon as the first one emptied protected them against running out of LPG while cooking. Those without a back-up LPG tank often reported switching to the fogón to finish cooking when their LPG tank ran out. Some women without a back-up tank reported that several days might pass before they could obtain an LPG refill. During this time, they usually cooked with the fogón.

“One tank is in the kitchen, and the other has to be full and ready. So when the one in the kitchen runs out, I have the spare one ready to change… I don’t have to be waiting for a new one… Also, this way I can use the voucher to exchange the tank as soon as it arrives [before the voucher expires].”

(Participant 8)

Having a backup tank was especially useful for women who lacked access to quick deliveries, and during holiday seasons when local distributors sometimes ran out of tanks due to high demand. The extra tank also provided time for women to save money for a refill. Additionally, women reported that sometimes LPG distributors seeking new customers or politicians seeking votes would come to the community offering discounted LPG tanks, and only those with an extra tank could take advantage of it.

“I would like to have two tanks. Then I would have gas while I am getting my money ready to buy more. Then I would have gas saved when I call them and ask them to deliver it. Then when I have money, I could fill the other one and save it. It would be good to have another one to use. It would be good that way, if you want to cook with only gas. Sometimes they come with an offer, to exchange a tank of gas for 8 or 10 soles, but my tank still has gas in it, so I miss out, because I don’t have an empty tank to exchange. For that reason, I’d like to have two tanks also.”

(Participant 19)

LPG affordability

Cost was commonly reported as a major barrier to sustained LPG use. In surveys, 53.5 % of participants from the intervention group reported ever participating in FISE at the end of the post-intervention period (month 24), increased from 40.5 % at the end of the intervention period in month 12. Despite FISE subsidies, however, 91.7 % (n = 79 out of 86) of participants from the intervention group reported that they perceived LPG to be expensive. Participants reported that on average, a full-price 10 kg tank of LPG cost 35.9 soles (10.9 USD) (SD 0.5, range 34–37), meaning a subsidized tank was around 20 soles (6.1 USD) (after applying the FISE subsidy of 16 soles [4.8 USD]). However, significantly more participants from the intervention group said they could afford to pay for two LPG tanks per month (17.4 %, n = 15) than from the control group (2.3 %, n = 2) at the end of the post-intervention period (chi-square p = 0.001).

All interviewees who received FISE benefits reported that the vouchers were vital to their ability to continue cooking with LPG. Some women described that the FISE vouchers made it feasible for them to purchase two LPG tanks per month (one at the discounted rate and one at full price), which was the average amount needed for exclusive LPG use (Williams et al., 2020c). Others said they could only afford one tank of LPG per month even with the discount, and thus continued to use the fogón for preparing foods that must cook for longer and animal food.

“I realized that gas is easier, so I have to get the money every 15 days… I use two tanks per month, one I buy using the voucher, the other with money, so I need to save money to buy a new one when it runs out… We sell milk and cheese… also eggs. We know that we need to save that money for the gas… It is easier with the voucher because the price is less.”

(Participant 6)

Several women mentioned that the FISE vouchers sometimes did not arrive on time and expressed concern that they could get cut from the program. When the voucher did not arrive, women said they either used their fogón, purchased an LPG tank at full price, or negotiated with the distributor to delay payment until the voucher arrived.

“The FISE voucher arrives every month, not every 15 days… But I use two gas tanks per month… And sometimes the voucher doesn’t arrive… or it arrives after two months… If I don’t have money, I cook with my fogón until my voucher arrives.”

(Participant 13)

Women who did not receive the FISE voucher described having more difficulty in purchasing LPG. While most continued to purchase some LPG at full price (typically no more than one tank per month), many reverted to cooking primarily with their fogón, reserving their LPG stove for secondary tasks such as preparing dinner and cooking during the rainy season or when in a rush. Some women said they had not enrolled in FISE because the enrollment process was difficult or they did not have time to travel to the nearby city to enroll.

“I used to cook for the pig and the dog on the gas stove, but now I cook for them on the fogon ´ to save gas… since it is no longer free, it costs me, and I have to save money… We don’t have [the voucher], so we have to pay 35 soles… Every month I buy gas for 35 soles. If I had a voucher, I would also cook for the pig and the dog [with gas]. With a voucher it would be cheaper.”

(Participant 10)

Overall, women said they continued to cook similar food during the post-intervention period (after they stopped receiving free LPG) as during the intervention, and thus had similar cooking energy demands. Many reported reverting to biomass for traditional dishes that tasted better when prepared on the traditional stove, dishes they had stopped preparing during the intervention period when fogón use was continuously monitored and discouraged. Additionally, some women reported adopting more LPG conservation strategies to increase affordability once they started paying for their own LPG. Strategies included cooking over a lower flame, enlisting other household members to help prepare ingredients and cook faster, peeling and cutting vegetables before lighting the stove, using a pressure cooker, cooking food together in one pot, soaking corn overnight before cooking, warming animal food for only a few minutes, wrapping food or pots in cloth to preserve warmth, and choosing foods that cook quickly. Some women also reported avoiding tasks that they had done when LPG was free, such as cooking lunch (eating cold leftovers instead), heating water for bathing or cleaning, toasting grains before grinding them into flour, cooking foods that take a long time to cook, and frequently preparing tea.

Discussion

Providing a free LPG stove, fuel, deliveries, and behavioral support for one year was successful in achieving a high rate of sustained LPG use after the intervention ended. While the intervention did not result in continued exclusive LPG use post-intervention, participants from the intervention group used LPG significantly more during the post-intervention period compared to control group participants in months 1–12 who owned LPG stoves but had not received such support. Participants from our intervention group also maintained a much higher degree of post-intervention LPG use than has been achieved by national LPG subsidization programs (Shankar et al., 2020). Qualitative data also indicate that intervention participants used LPG significantly more after the intervention compared to their pre-intervention cooking practices.

Women’s experience using LPG exclusively and observing its benefits firsthand during the intervention was a key motivator of continued use post-intervention. Additionally, the project strove to remove all barriers to LPG use during the trial, by providing free fuel (removing financial barriers), delivering it directly to households (removing barriers related to transportation, travel time, and delivery fees), teaching women about safe LPG use (reducing barriers related to knowledge and self-efficacy), responding quickly to problems and advising women about why the problem occurred and how to address it (reducing fears), and providing a high quality stove and tanks (reducing development of negative opinions about LPG that could arise from poor quality equipment). These efforts instilled a positive impression of LPG, motivating participants to continue using it despite post-intervention barriers. Participants may not have been willing to overcome these challenges had they not become accustomed to LPG and its benefits during the intervention period.

A facilitating environment, in which consumers have access to multiple LPG distributors motivated by competition to offer benefits, good service, affordable delivery, and a stable LPG supply, is also key to continued LPG use. Previous studies have similarly shown that large distances to and unavailability of LPG distributors are key barriers to continued LPG use (Asante et al., 2018; Puzzolo et al., 2016). When women can be sure of obtaining LPG at an expected price, they may feel more confident in removing their traditional stove, thus reducing the temptation and likelihood of reverting to biomass (Williams et al., 2020a). However, the perceived unpredictability of the FISE benefits, such as concerns about getting cut or vouchers not arriving, may hamper their willingness to remove their biomass stoves. Clarity to consumers about what circumstances would cause someone to be removed from the program and improvements to the system to ensure that vouchers arrive consistently each month could help increase confidence in the program and motivate more women to remove their biomass stove. Streamlining the process of enrolling in FISE in the first place could also facilitate greater participation and enhance LPG affordability in more households, as was similarly recommended by Pollard et al. (2018).

Our findings also highlight several other strategies that could be employed within LPG promotion efforts to encourage greater sustained use of LPG. Training delivery personnel to check the functionality of LPG stoves and tanks could allay fears and encourage greater use. Providing toll free lines for LPG delivery requests and establishing credit systems for LPG purchases could also facilitate more consistent use. Ensuring that women have access to a spare LPG tank or instituting a pay-as-you-go system in which customers could pay electronically in small increments for LPG as it is used could help remove gaps in LPG supply and enable more continuous use (Shupler et al., 2021). Some monitoring and reinforcement of fogón abandonment by LPG providers could also help reduce stacking.

Results from our research also indicate that the intervention may have increased willingness to pay for LPG among participants in the intervention group following completion of the intervention. Most intervention group participants reported buying at least one tank of LPG per month in the post-intervention period, which previous research suggests would cover approximately 50 % of cooking needs (Williams et al., 2020c). Many participants purchased additional tanks beyond those subsidized by the FISE program, indicating that they were willing to pay more for LPG, but unable to consistently buy a full-price tank of LPG every month in addition to the discounted tank. Some additional financial support above the FISE subsidy, coupled with training on LPG conservation strategies, could be enough to enable exclusive LPG use. Identifying other methods for completing tasks commonly done with the fogón, such as thermal bags for dishes that need to boil for long periods or low-cost solar cookers for preparing animal food, could help reduce fuel demands and enhance the affordability of exclusive clean fuel use.

Strengths of our study include triangulation between several sources of quantitative and qualitative data, development of trust between participants and fieldworkers stimulating honesty and openness in responses, and high participant retention and data completeness. Limitations include gaps in stove use monitoring when fogones were re-installed prior to fieldworkers observing and installing a SUMs in them, and potential cross-over in terms of SUMs on fogones picking up LPG cooking events and vice versa when stoves were located close together.

Conclusion

Our study indicates that an initial period of higher financial support followed by a lower subsidy could stimulate greater LPG adoption than has previously been achieved in national LPG programs. A short period of greater financial support may be more financially feasible for governments than indefinitely providing an increased subsidy. However, additional efforts beyond initial intensive support will be necessary to achieve exclusive LPG use, as well as infrastructural development in areas currently lacking LPG access to foster an enabling environment for LPG adoption and sustained use.

Appendix A

Cardiopulmonary outcomes and Household Air Pollution (CHAP) trial Investigators: Steering Committee: William Checkley MD PhD (Johns Hopkins University, Baltimore, MD, USA), Gustavo F Gonzales MD (Universidad Peruana Cayetano Heredia, Lima, Peru), Luke Naeher PhD (University of Georgia, Athens, GA, USA), Joshua Rosenthal PhD (National Institutes of Health, Bethesda, MD, USA), N Kyle Steenland PhD (Emory University, Atlanta, Georgia, USA). Johns Hopkins University Investigators: Theresa Aguilar, Vanessa Burrowes PhD, Magdalena Fandiño-Del-Rio PhD, Elizabeth C Fung MSPH, Dina Goodman MSPH, Steven A Harvey PhD, Phabiola Herrera MD, Josiah L Kephart PhD, Kirsten Koehler PhD, Alexander Lee, Kathryn A Lee MPH, Catherine H Miele MD MPH, Mitra Moazzami MSPH, Lawrence Moulton PhD, Saachi Nangia, Laura Nicolaou PhD, Carolyn J O’Brien MSPH, Timothy R Shade, Suzanne Simkovich MD MS, Lena Stashko MSPH, Ariadne Villegas-Gomez MSPH, Kendra N Williams PhD, Abigail Winiker MSPH. Asociación Benéfica PRISMA Investigators: Marilu Chiang MD MPH, Gary Malpartida, Carla Tarazona-Meza MPH. Washington University Investigators: Victor Davila-Roman MD, Lisa de las Fuentes MD. Emory University Investigators: Dana Barr Boyd PhD, Maria Jolly MSPH, Angela Rozo MS.