Background: Three billion individuals worldwide rely on biomass fuel [dung, wood, crops] for cooking and heating. Further, health conditions resulting from household air pollution (HAP) are responsible for approximately 3.9 million premature deaths each year. Though transition away from traditional biomass stoves is projected curb the health effects of HAP by mitigating exposure, the benefits of newer clean cookstove technologies can only be fully realized if use of these new stoves is exclusive and sustained. However, the conditions under which individuals adopt and sustain use of clean cookstoves is not well understood. Methods: The Enhancing LPG Adoption in Ghana (ELAG) study is a cluster-randomized controlled trial employing a factorial intervention design. The first component is a behavior change intervention based on the Risks, Attitudes, Norms, Abilities, and Self-regulation (RANAS) model. This intervention seeks to align these five behavioral factors with clean cookstove adoption and sustained use. A second intervention is access-related and will improve LPG availability by offering a direct-delivery refueling service. These two interventions will be integrated via a factorial design whereby 27 communities are assigned to one of the following: the control arm, the educational intervention, the delivery, or a combined intervention. Intervention allocation is determined by a covariate-constrained randomization approach. After intervention, approximately 900 households’ individual fuel use is tracked for 12 months via iButton stove use monitors. Analysis will include hierarchical linear models used to compare intervention households’ fuel use to control households. Discussion: Literature to-date demonstrates that recipients of improved cookstoves rarely completely adopt the new technology. Instead, they often practice partial adoption (fuel stacking). Consequently, interventions are needed to influence adoption patterns and simultaneously to understand drivers of fuel adoption. Ensuring uptake, adoption, and sustained use of improved cookstove technologies can then lead to HAP- reductions and consequent improvements in public health. Trial registration: NCT03352830 (November 24, 2017). Keywords: Clean cookstoves, Household air pollution, Sustained use, Clean cookstove adoption, Behavioral intervention, Structural intervention, Biomass combustion * Correspondence: email@example.com Department of Environmental Health Sciences, Mailman School of Public Health, Columbia University, 722 W. 168th Street – 11th Floor, New York, NY 10032, USA Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Carrión et al. BMC Public Health (2018) 18:689 Page 2 of 11 Background with adoption. These variables can be broadly divided Three billion people rely on biomass fuels for their cook- into three categories: household/setting characteristics, ing and heating needs worldwide. Biomass fuels consist infrastructure, and knowledge and perceptions [20–22]. of dung, wood, charcoal, crop waste, etc. . Combus- Household/setting characteristics are features that de- tion of these fuels leads to high levels of particulate mat- scribe the household, their neighbors, and/or communi- ter, carbon monoxide, polycyclic aromatic hydrocarbons, ties. Examples include ethnicity, religion, maternal and and other deleterious air pollutants [2–6]. In this con- paternal education, female head of household, parental text, these air pollutants are collectively referred to as and family wealth/income, household size, and age. household air pollution (HAP). The health effects of While it is regularly true that these variables predict HAP are vast and wide-ranging. In fact, it is estimated adoption, the direction of the association varies across that 3.9 million premature deaths are attributable to studies . The reasons behind these inconsistencies HAP annually . Deaths attributable to HAP occur are unexplained, but may represent uncontrolled con- from a diverse set of diseases such as stroke, ischemic founding based on underlying constructs that are con- heart disease, pneumonia, cataracts, etc. [3, 7]. Although textually relevant, or they may indicate the central role disease etiology and causal mechanisms are outstanding, of local conditions in shaping decision processes. it is widely acknowledged that HAP is a severe health Access-related factors associated with cookstove adop- threat. Therefore, efforts are being made to characterize tion include 1) financial, tax, and subsidy aspects (2) and mitigate exposure. This protocol paper outlines the market development (3) regulation, legislation, and stan- Enhancing LPG Adoption in Ghana (ELAG) study, a dards, and (4) programmatic and policy mechanisms cluster-randomized controlled trial designed to increase [20–22]. Broadly defined, the access factors outline con- LPG stove adoption and sustained use. textual physical and/or organizational facilitators of clean cookstove adoption and sustained use. Fuel access Mitigating HAP exposures factors are also oftentimes specific to the stove type. For HAP exposures result from combustion of biomass and example, improved biomass cookstoves necessitate a dif- other solid fuels in traditional cookstoves . Public ferent fuel infrastructure than LPG stoves. After an ini- health proponents have looked to cleaner cookstoves to tial stove purchase, some stoves require repeated reduce HAP exposures [9, 10]. However, a number of re- purchase of fuels. Users are then responsive to the price cent studies establish that clean cookstoves do not auto- of the physical stove, but also fuel prices. Considering matically reduce HAP exposures . Stove stacking, the fuel access environment of cookstove adoption is im- wherein households partially adopt the new technology perative for HAP-related interventions. while maintaining use of traditional cooking technolo- Understanding knowledge and perceptions preventing gies, is a core challenge [12–15]. While partial adoption behavior change is vital to any health-related interven- of cleaner cookstoves may partially reduce exposure, tion. Studies have shown numerous associations with prior work has shown that these reductions are not suf- cookstove adoption, including knowledge/perceptions ficient to eliminate risk . Another challenge facing of: the health impacts of HAP, safety benefits of new clean cookstove interventions is that community-level cookstoves, time-savings benefits, improved cleanliness emissions may substantially contribute to individuals’ ex- of newer stoves, social norms, newer cookstove users posure [17, 18]. Therefore, large scale adoption of clean within a social circle, and the cultural appropriateness of cooking technologies may be required to decrease over- technologies [20, 21]. Generally speaking, knowledge is all HAP exposures. These challenges have led to a body regarded as highly modifiable whereas attitudes can be of scientific literature dedicated to clean cookstove adop- more challenging to alter [23, 24]. Both elements, how- tion that addresses the determinants of uptake of the ever, must be aligned with a new behavior in order to new technologies [19–22]. Ultimately these studies seek observe behavior change . to understand opportunities to intervene on HAP expos- ure. For our purposes, ‘improved cookstoves’ refer to Challenges in cookstove adoption research those which still utilize biomass, but increase the effi- Studying cookstove adoption is both conceptually and ciency of combustion and thereby reduce HAP expo- methodologically challenging. Disciplines involved span sures. ‘Clean cookstoves’ on the other hand, refers to the social, environmental, and health sciences, utilizing non-biomass stoves, including liquefied petroleum gas quantitative, qualitative, and mixed methods [26–30]. (LPG), induction, solar, biogas, etc. The field has largely employed observational study designs to date. While these studies are quite inform- Efficient cookstove adoption ative, they may be vulnerable to selection bias because The scientific literature regarding cookstove adoption individuals opt into each group of the study by deciding has identified numerous pre-conditions often associated whether or not to purchase a stove, sustain use, etc. Carrión et al. BMC Public Health (2018) 18:689 Page 3 of 11 There may be underlying characteristics that predict this study will offer novel insights into the predictors of entry into each group, thus limiting the generalizability sustained use, strategies that can be employed to in- of findings. Controlled trials can address these limita- crease use, and important policy actions that can reduce tions through the randomization and follow up of partic- exposures to HAP and its health consequences. ipants, but there are few studies utilizing these study designs [31, 32]. Methods/design Most cookstove adoption studies have focused on ini- This study builds on an ongoing successful collabor- tial adoption versus sustained use. [21, 33]. This is an ation between Columbia University in the City of important distinction because adoption studies have New York and the Kintampo Health Research Centre largely focused on the enablers and barriers of initial (KHRC). In fact, the study is an outgrowth of the stove acquisition and/or the use of the technology early Ghana Randomized Air Pollution and Health Study in its adoption . However, there are many reasons to (GRAPHS), which was a 5-year cluster-randomized believe that behaviors change over time. For example, controlled trial assessing the impacts of a HAP researchers have noted situations wherein new stove use intervention on low birthweight and pneumonia . is high upon acquisition, but decreases over time. There GRAPHS included one control, and two intervention are plausible reasons why participants would decrease arms. LPG stove users and improved cookstove (Bio- use. New stoves could break with consistent use, and lite) users served as the interventions and the trad- without access or means of repair, participants would itional 3-stone fire users were the control arm. likely default to the traditional stove. It is also possible Ethical considerations dictated that the control arm that a household’s financial circumstances change and would receive clean cookstoves upon study comple- use falters. Without clear plans to recover from these ex- tion since LPG stoves are believed to reduce HAP ex- ternal stimuli, users would resume traditional stove use. posures the most substantially. Given remaining While there is a small number of studies focused on sus- resources upon study completion, all participants in tained use, that amount is growing. This is because re- the control and Biolite arms were scheduled to re- searchers increasingly recognize the importance and ceive LPG stoves at study closeout. This provided an complexity of the issue [13, 34, 35]. opportunity to assess patterns of adoption and sus- Household/setting characteristics, infrastructure, and tained use of LPG cookstoves among a large group of knowledge and perceptions are all highly contextual is- participants. sues. Although adoption has been extensively studied around the world, much work remains. Sub-Saharan Af- rica has the largest proportion of individuals using bio- Hypothesis mass fuels for cooking, and is the only region globally We hypothesize that households that receive both the where traditional biomass use is still growing . Sus- behavioral and access interventions will demonstrate tained use studies are small in number and limited geo- higher levels of sustained use in the last 6-months of graphically. To our knowledge, there have been few the study period, compared to those in the no studies in sub-Saharan Africa, demonstrating a need for intervention group, see Table 1. continued research in an important region. The goal of this paper is to outline the study design for the ELAG Study. The objectives of ELAG is to assess Study setting the effectiveness of two interventions on facilitating sus- The study area is in Kintampo North Municipality tained use of LPG. A cluster-randomized trial with a fac- and Kintampo South District in the Brong-Ahafo torial design is being used to test the effectiveness of Region of Ghana. This is a mostly rural area (popu- two distinct interventions: 1) a behavioral change inter- lation 176,480), see Fig. 1 . Households in the vention using the Risks, Attitudes, Norms, Abilities, and study area traditionally use three-stone fires for their Self-Maintenance (RANAS) model and (2) an access cooking needs. Ghana has a warm climate, with an intervention to modify the ease of refueling . The annual average temperature is 26 °C . Therefore, factorial design also allows us to evaluate the interaction stoves are typically only used for cooking, not heat- of these two interventions. We deliver these interven- ing. There are two seasons, wet and dry. During the tions to mothers and, when possible, their partners. Sus- dry season most cooking takes place outdoors while tained use is measured with stove use monitors (SUMs). enclosed or covered kitchen areas are the site of These monitors will be in place for 12 months after most cooking in the wet season. Wood is the main intervention delivery. Sustained use will be assessed by fuel source in the study area, but charcoal is used as analyzing the effect of the interventions on stove use in well . the last 6 months of the study period. We believe that Carrión et al. BMC Public Health (2018) 18:689 Page 4 of 11 Table 1 Allocation of clusters by study arm (and number of households) No educational intervention Educational Intervention Totals No. of communities (households) No. of communities (households) No agent delivery 7 (271) 7 (243) 14 (514) Agent delivery 7 (241) 6 (224) 13 (465) Totals 13 (492) 12 (451) 27 (979) Study eligibility & recruitment KHRC initiatives, we do not anticipate any issues ELAG participants are limited to participants who reaching our recruitment goals. were enrolled in the original GRAPHS cohort and who: 1) were originally randomized to the Biolite or Ethics approval and consent control arms of the study and 2) still reside in the This study has received approval from the Institutional KHRC study region (see Fig. 1). Participants are vis- Review Board of Columbia University Medical Center and ited to assess interest in the new study, although, due the Kintampo Heath Research Centre Institutional Ethics to longstanding involvement with GRAPHS and other Committee. The study is registered with clinicaltrials.gov Fig. 1 Map of Kintampo, the study area Carrión et al. BMC Public Health (2018) 18:689 Page 5 of 11 under NCT03352830. Informed consent is obtained by that underpins the RANAS model is that each of these KHRC fieldworkers from all ELAG study participants five behavioral factors are necessary, but not sufficient, prior to enrollment. to induce behavior change. After baseline data collec- tion, ELAG households will be convened in cluster-wide Power meetings for LPG stove distribution and the behavioral Power calculations reflect the cluster-randomized study change intervention. A research team member and a design. Treatment is on the cluster level (2) while out- peer-adopter will collaborate to deliver the intervention. comes are on the household level (1). The outcome of The peer-adopter is a participant from a GRAPHS LPG the study is minutes of LPG use per day summed over a community who has maintained use of LPG after study six-month period. Multilevel model power calculations conclusion. ELAG will rigorously employ RANAS to with 27 clusters, 979 sample size and type I error at 5%. promote clean cookstove adoption, see Additional file 1 The mean for the reference group is estimated at for the scripts used to guide the interventions. 3000 min over 6 months, with equal group sizes. This is Like other behavioral change campaigns, RANAS in- believed to be a conservative estimate based on our un- volves communicating risks associated with the trad- published research in the region, which found ~ 3100 itional behavior. Not only should participants be aware mean minutes of use over 20 weeks . The effect size of the severity of risks, but they should be made to (Cohen’s D) is calculated with 1000 min as the pooled recognize that they are vulnerable to those potential standard deviation, which is an overestimation based on health outcomes [25, 44, 45]. Participants are first intro- our research. Multilevel power formulas were derived duced to the concept of HAP. Then they are presented from Scherbaum, 2009 . Intraclass Correlation Coef- with a series of pictures that show adverse outcomes ficients are unknowable given the novelty of the that have been shown to be caused by HAP, and that are research, but several possibilities were modeled - visual- recognizable to community members (cataracts, low ized with the Optimal Design Plus Empirical Evidence birth weight, and respiratory diseases) [3, 7]. Addition- version 3.1 software, see Fig. 2. ally, pictures of blackened kitchen walls are used to ex- plain that the same pollutants which dirty the walls also Health promotion intervention enter human lungs when exposed to smoke. The study consists of two types of interventions, a be- Attitudinal factors include perceptions of time, money, havioral change intervention and an access intervention. and effort associated with the behavior change, and the Our team selected the Risks, Attitudes, Norms, Abilli- benefits of the new behavior. Following the presentation ties, and Self-Maintenance (RANAS) model to design a of risks, participants will be informed of the potential clean cookstove behavioral change intervention . health benefits of reduced HAP exposure. Other benefits The RANAS model was originally designed and will be presented, such as reduced time dedicated to employed for issues pertaining to water, sanitation, and wood collection which can then be reallocated for edu- hygiene. However, we recognized its potential applica- cational or economic goals . tion for clean cookstove adoption. The core assumption Normative factors describe perceived expectations from peers, leaders, and/or of one’s self. At this point the LPG peer adopter provides a testimonial regarding their experiences using LPG and overall appreciation of the technology. This is complimented with the a re- search staff explaining the Ghanaian government’s ef- forts to reduce HAP-exposures and protect Ghanaian natural resources from deforestation . The commu- nity is then prompted to make a collective commitment to using LPG. The ability-related behavioral factor represents the partic- ipants’ confidence in performing the new behavior. These abilities include safe, effective, and culturally-appropriate useofanLPG stove, whichmay requiresomeorientation. Indeed there is some literature that documents new user’s reticence to use a new stove because it is unclear how cul- turallysuitableitis[48, 49]. Therefore, it is important to demonstrate how to cook traditional Ghanaian meals on the LPG stoves. An LPG peer adopter will have a food Fig. 2 Power Calculations at .3 and .5 effect sizes demonstration where they cook a traditional meal for the Carrión et al. BMC Public Health (2018) 18:689 Page 6 of 11 participants in that cluster. On a longer timeframe, individ- of conditions. The ‘No educational intervention’ and ‘No uals may be unsure how to sustain the repeated payments agent delivery’ will serve as the control group, while all required to refill the LPG cylinder [48, 50]. The interven- other permutations of the two interventions will be tion will also include a financial literacy orientation, which compared to the control group during the analysis will provide strategies on how to smooth payments over phase. As per the parent study (GRAPHS), this is a clus- time, including savings strategies and credit. A key compo- ter randomized control trial. Randomization will occur nent of the orientation will be a discussion among partici- on level 2 (villages) while outcomes will be measured at pants in smaller groups, on different scenarios of level 1 (households). household financial limitations and the different strategies they would use to meet the financial obligation of using Randomization LPG stoves. Participants will also be provided with susu The parent study utilized a cluster-randomized design boxes (savings boxes) and encouraged to make weekly de- on coarsened exact matches to assign study treatments posits towards the refilling of LPG cylinders. . This study re-randomized clusters employing a Self-regulation factors provide continued orientation covariate-constrained randomization approach with sev- to the desired behavior in anticipation of conflicts or eral identified prognostic covariates [55, 56]. The distracting cues to the old behavior. The self-regulation covariate-constrained randomization approach is a factor is designed with the assumption that relapse to the powerful allocation technique to ensure balance between old behavior is inevitable for many individuals because arms in cluster randomized trials. Baseline covariates outside circumstances cannot be controlled. ELAG ad- were chosen based on prior literature and theoretical re- dresses this factor by contracting and training lationships (see Table 2). When covariates are continu- Community-based Surveillance Volunteers (CBSVs) who ous, balance is determined via mean differences between will visit participants weekly to discuss the potential bar- treatment arms. Maximum permissible imbalance is des- riers to sustained use and brainstorm possible solutions. ignated a priori. Given these parameters, allocation This is to assist sustained use efforts throughout the study amongst arms is randomly designated if below the max- period. CBSVs are trained to recognize and assist with is- imum permissible imbalance. An independent epidemi- sues that would interfere with LPG use, and serving over- ologist performed the final randomization using the all as a resource for households. CBSVs have a ccrand procedure in Stata . Allocation was not re- long-standing presence in the communities, charged with vealed to field staff until after baseline data collection tracking health and demographics in each community. was complete. Access intervention Stopping guidelines Policy makers and researchers have suggested the im- Positive stopping rule portance of the ‘last mile’ (or, more realistically in rural Due to the nature of the intervention, the outlined study Ghana, last 30 km ) of LPG delivery/accessibility, does not have a positive stopping rule. The hypothesis is but few have empirically demonstrated the degree to that an educational intervention and an agent-delivery which logistical barriers impact user demand [51–53]. system will meaningfully increase sustained use of the This intervention will determine the degree to which clean cookstoves. Identifying the significance and magni- physical accessibility of a product influences sustained tude of that improvement is a policy-relevant endeavor use. Each community has at least one taxi driver or as it affords the opportunity to conduct cost-benefit ana- motor-king (tricycle cart) rider who can be paid to trans- lyses of expanded implementation. Our current outcome port individuals or goods. KHRC will contract such measure is time-dependent, wherein we will compare drivers/riders in each community to provide pickup and cookstove usage over the 12 months of the intervention. drop off services for LPG cylinders. For this study, households in access intervention clusters will bear the Negative stopping rule cost of the refilling while the cost of transportation will This is a behavioral intervention designed to determine be paid for with vouchers provided by the study. This effective strategies to increase the uptake of clean ener- intervention addresses an important gap in our know- gies. The control group is a ‘business-as-usual’ approach, ledge surrounding sustained demand and use patterns of and we hypothesize the intervention groups will improve LPG fuels. their uptake of the new technologies. We plan to halt study activities if 1) there are any LPG cookstove-related Factorial design accidents that occur within our cohort and the study Each cell of the factorial design will be functionally timeframe that result in permanent bodily injury or treated as an ‘arm’ of the study, see Table 1. This means death, or (2) if the intervention groups show statistically that no one group will experience exactly the same set significant decreases in LPG cookstove usage. Carrión et al. BMC Public Health (2018) 18:689 Page 7 of 11 Table 2 Covariate constrained randomization variables Variable Rationale Community Asset Index Scientific articles have shown that differential access to resources can be predictive in the uptake of new cookstove technologies (19–21). Average Household Size Scientific articles have shown that household size can be predictive in the uptake of new cookstove technologies (19–21). Distance to Refueling Study communities are scattered throughout the region at varying distances from the refueling center. Further distances Station are likely a deterrent to refuel for non-Agent delivery households, see map. Households per cluster To ensure roughly equal number of participating households per arm. Baseline data collection and covariates Fuel wood is often free to the household because it sim- Information on several constructs will be collected be- ply requires that a family member spend time to collect fore delivering the behavioral and infrastructural inter- material in the local environs or farms. Increasing LPG ventions. Beyond standard baseline demographic and prices may lead some families to default to wood use socioeconomic status surveys, a pre/post-test of RANAS [20, 21]. The research team will track fuel prices by call- model behavioral factors will be administered in order to ing or visiting the LPG refueling station on a regular document any changes in participants knowledge, per- basis. This will allow us to assess individual responsive- ceptions, or attitudes regarding HAP and/or LPG stoves ness to a dynamic fuel landscape. [25, 44, 45]. This will provide vital insight on the role of household/setting characteristics on sustained use, and Outcome measures the effectiveness of the various components of the be- The principal outcome of interest is minutes of LPG havioral intervention on knowledge/perceptions of HAP stoveuse over thelast6months of thestudy.This and cookstoves. time period is of interest in order to assess the effect- Many studies have shown that gender can play an im- iveness of the intervention on sustained use rather portant role in predicting cookstove adoption [58, 59]. than initial adoption. Stove use will be measured via Intra-household cooperation is a conceptual framework stove use monitors (SUMs). Each stove will be that makes explicit the bargaining process of equipped with SUMs, which are iButton temperature decision-making within a household. Contrary to earlier loggers programmed to collect temperature data at 10 economic theory, households do not operate as a unit. minute intervals . This leaves the memory at cap- Instead, several actors negotiate decision-making, and acity after 2 weeks. Field staff will visit households power is rarely symmetrically distributed within the every 2 weeks to download the data. Monitors will be household [60, 61]. This asymmetry tends to follow gen- used to determine minutes of stove use during that dered power dynamics. The concept of intra-household period. Biweekly visits will also serve as a quality con- bargaining could thus mediate the relationship between trol measure because staff can address faulty readings, the intervention and increased sustained use. Our study typically by replacing the iButton. will utilize a modified version of the dictator game to as- A secondary outcome of interest is stove use mea- sess intra-household cooperation [62, 63]. The game re- sured via weighing of LPG cylinders with a scale. quires each partner of the household be separated Field staff are scheduled to visit households every 2 temporarily for administration. Once separate, they are weeks to download SUMS data. They will use this op- asked to select one of three envelopes with different portunity to administer an LPG stove use question- pre-portioned sums of money. They must then decide naire and weigh cylinders with a scale during their whether to keep the entire sum of money for themselves, bi-weekly visits. Weighing cylinders is a cheaper alter- or direct all or a portion of it to their spouse and/or native to measuring stove use, but it is possible that charity. The player can chose to send a certain amount the measurements are biased. When a cylinder is of money to the third party. The money has a 50/50 weighed, participants will be asked if they refilled the chance of reaching that party and, if so, being doubled. cylinder within the 2 weeks and, if so, how much. It In a perfectly cooperative household, it is within each in- is possible that social-desirability bias is introduced in dividual’s interest to send all of the money to their part- these measurements because participants are aware of ner. Intra-household cooperation, then, is measured as our study interests and may misreport accordingly the ratio of money sent between partners over the entire . Determining the degree of bias or measurement amount received. Note that payout is hypothetical and error by comparison to SUMS data may be useful for currency will not actually exchange hands. future studies requiring stove use measurements in LPG fuel prices are a fluctuating covariate of interest, Ghana or similar contexts. See Fig. 3 for a study representing a potential access barrier to sustained use. timeline. Carrión et al. BMC Public Health (2018) 18:689 Page 8 of 11 Data management the only individuals permitted to access the data are those Field staff will administer numerous paper based surveys, listed in the Columbia University IRB protocol. Standard and survey forms will be checked for completeness and KHRC and Columbia University procedures will be consistency by a field supervisor. Data entry will be per- followed otherwise. Monthly reports providing summaries formed in the KHRC computer center. KHRC has of enrollment and various field activities are discussed well-established procedures for digitizing data and storing among the study team. results in a relational database. The original paper records will be securely stored in the KHRC Data Center. Once Analysis paper records are transcribed, data will be anonymized for The primary analysis will be to examine the effect of the privacy protection. Personally identifiable information will intervention on average stove use over the final 6 be removed and a proprietary KHRC participant ID will months of the observation period. Due to the nested na- link records. Clinical health information is not included in ture of the observations, these data will be analyzed with or collected by this particular study. Outside of KHRC, a hierarchical linear model. The three intervention arms, Fig. 3 ELAG Study Flowchart Carrión et al. BMC Public Health (2018) 18:689 Page 9 of 11 consisting of the two interventions alone and one com- Additional file bined, will be compared to the control arm, which re- Additional file 1: Script for delivering educational messages to ceives no intervention. Adoption Aim 3 Participants. (DOCX 26 kb) Several secondary analyses are also of interest. First, while the primary analysis is unadjusted for potential co- Abbreviations variates, a secondary analysis will include covariates such CBSV: Community-based surveillance volunteer; ELAG: Encouraging LPG as financial literacy, the number of unanticipated life Adoption in Ghana study; GRAPHS: Ghana Randomized Air Pollution and Health Study; HAP: Household Air Pollution; KHRC: Kintampo Health Research events necessitating additional financial resources, ethni- Centre; LPG: Liquefied petroleum gas; RANAS: Risks, Attitudes, Norms, city, intra-household cooperation, and household asset Abilities, and Self-regulation; SUMs: Stove use monitors index. Second, additional analyses will look at the vari- ous trends across time, assessing stove use in the first 5 Acknowledgments Thanks to Kenneth Wiru, who produced the map of the study region months of the study and over the 10 months. A third depicted in Fig. 1. analysis will analyze sustained use across arms by demo- graphic characteristics identified in improved cookstove Funding This study is funded by the National Institute of Environmental Health adoption related systematic reviews. Finally, given that Sciences [NIEHS] under grant number R01ES024489. In addition, Daniel the Ghanaian government is already distributing LPG Carrión is currently supported under NIEHS Grant T32ES023770. cookstoves and interested in increasing sustained use, Authors’ contributions this research is policy-relevant. Qualitative data will be DC reviewed relevant background literature, designed the behavior change collected and analyzed among key stakeholders involved intervention, conducted power analyses, drafted and edited baseline surveys, in the intervention. This analysis can then be used to collaborated on overall study design and integration, and wrote and edited the manuscript. DJ and KPA conceived of the overall study design, consider meaningful ways to scale the intervention based collaborated on both behavioral and access intervention designs, co-led the on existing fuel infrastructure. parent study, edited the manuscript, and serve as principal investigators of the NIEHS grant. RD coordinates field activities, drafted and edited surveys, edited the manuscript, and participated in intervention design. FA led quali- tative data collection for the pilot study, collaborated on final intervention Discussion details after reviewing pilot results, and participated in survey design and ELAG represents a unique opportunity to understand edits. OA oversees contributes to stove use monitor data design and data various facets of LPG adoption and sustained use, collection, quality assurance and improvement. MT and OA designed the data management system of the trial, including survey transcription, principally the role of two distinct interventions to in- digitization, database development, and report development. TT participated fluence ongoing use, one via behavioral change pro- in survey design and edits, and collaborated on intervention design. MM par- motion and another through fuel-access ticipates in coordination of field activities and data collection. All authors read and approved the final manuscript. modifications. However, the large sample size, ran- domized design, and established cohort, allows for ex- Ethics approval and consent to participate ploration of numerous other pivotal relationships. For Individuals from the parent study, GRAPHS, will be directly contacted and asked to enroll in the current study. Participation will be confirmed through example, assessing the role of gender via an informed consent process, which will be documented and consent can intra-household cooperation as a potential mediator be revoked at any time. This study has received approval from the of increasing sustained use is both innovative and in- Institutional Review Board of Columbia University Medical Center and the Kintampo Heath Research Centre Institutional Ethics Committee. Additional structive. Quantifying the degree to which unexpected approvals will be sought if there are unanticipated modifications to the life circumstances impedes sustained use is also novel protocol. and useful to policymakers. Competing interests Biomass cookstove use is a widespread source of The authors declare that they do not have any competing interests. air pollution, mostly in the developing world. How- ever, much remains unknown regarding effective Publisher’sNote strategies to increase clean cookstove adoption and Springer Nature remains neutral with regard to jurisdictional claims in sustained use. Many studies to date have sought to published maps and institutional affiliations. demonstrate the health benefits of improved or clean Author details cookstove, but have been largely unsuccessful, likely Department of Environmental Health Sciences, Mailman School of Public due to the difficulties of lowering exposures to Health, Columbia University, 722 W. 168th Street – 11th Floor, New York, NY 10032, USA. Kintampo Health Research Centre, Kintampo, Ghana. health-relevant levels. Indeed, many studies have shown that exposure-reduction is a challenging en- Received: 14 March 2018 Accepted: 28 May 2018 deavor . Our study aims to document meaningful ways to increase adoption and sustained use, thereby References reducing exposure. If successful, these strategies can 1. Bonjour S, Adair-Rohani H, Wolf J, Bruce NG, Mehta S, Pruss-Ustun A, et al. then be scaled to measure the health impacts of ex- Solid fuel use for household cooking: country and regional estimates for posure reductions. 1980-2010. Env Health Perspect. 2013;121:784–90. Carrión et al. BMC Public Health (2018) 18:689 Page 10 of 11 2. Fullerton DG, Bruce N, Gordon SB. Indoor air pollution from biomass fuel 25. Contzen N, Mosler HJ. The RANAS approach to systematic behavior change. smoke is a major health concern in the developing world. Trans R Soc Trop Methodological Fact Sheet 1 [Internet]. Dübendorf, Switzerland: Eawag; Med Hyg. 2008;102:843–51. 2015. Available from: https://www.eawag.ch/fileadmin/Domain1/ 3. Smith KR, Bruce N, Balakrishnan K, Adair-Rohani H, Balmes J, Chafe Z, et al. Abteilungen/ess/schwerpunkte/ehpsy/RANAS_Methodological_Fact_Sheets. Millions dead: how do we know and what does it mean? Methods used in pdf. Accessed 13 June 2017. the comparative risk assessment of household air pollution. Annu Rev 26. Stanistreet D, Hyseni L, Bashin M, Sadumah I, Pope D, Sage M, et al. The role Public Health. 2014;35:185–206. of mixed methods in improved cookstove research. J Health Commun. 4. Streets DG, Bond TC, Carmichael GR, Fernandes SD, Fu Q, He D, et al. An 2015;20(Suppl 1):84–93. inventory of gaseous and primary aerosol emissions in Asia in the year 27. Bhojvaid V, Jeuland M, Kar A, Lewis JJ, Pattanayak SK, Ramanathan N, et al. 2000. J Geophys Res Atmospheres. 2003;108(D21):n/a. How do people in rural India perceive improved stoves and clean fuel? 5. Gustafsson Ö, Kruså M, Zencak Z, Sheesley RJ, Granat L, Engström E, et al. Evidence from Uttar Pradesh and Uttarakhand. Int J Environ Res Public Brown clouds over South Asia: biomass or fossil fuel combustion? Science. Health. 2014 Feb;11(2):1341–58. 2009;323(5913):495. 28. Hankey S, Sullivan K, Kinnick A, Koskey A, Grande K, Davidson JH, et al. 6. Christian TJ. Trace gas and particle emissions from domestic and industrial Using objective measures of stove use and indoor air quality to evaluate a biofuel use and garbage burning in Central Mexico. Atmospheric Chem cookstove intervention in rural Uganda. Energy Sustain Dev. 2015 Apr;25: Phys. 2010;10(2):565–84. 67–74. 7. Lim SS, Vos T, Flaxman AD, Danaei G, Shibuya K, Adair-Rohani H, et al. A 29. Alem Y, Beyene AD, Köhlin G, Mekonnen A. Modeling household cooking comparative risk assessment of burden of disease and injury attributable to fuel choice: a panel multinomial logit approach. Energy Econ. 2016;59: 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a 129–37. systematic analysis for the global burden of disease study 2010. Lancet. 30. Jeuland M, Pattanayak SK, Soo T, Sheng J. Preference Heterogeneity and 2013;380:2224–60. Adoption of Environmental Health Improvements: Evidence from a Cookstove Promotion Experiment [Internet]. Rochester, NY: Social Science 8. Kim Oanh NT, Reutergårdh LB, Dung NT. Emission of polycyclic aromatic Research Network; 2014. Available from: https://pdf.usaid.gov/pdf_docs/ hydrocarbons and particulate matter from domestic combustion of selected pbaab016.pdf. Accessed 21 Jan 2017. fuels. Environ Sci Technol. 1999;33(16):2703–9. 9. Mortimer K, Ndamala CB, Naunje AW, Malava J, Katundu C, Weston W, et al. 31. Beltramo T, Blalock G, Levine DI, Simons AM. Does peer use influence A cleaner burning biomass-fuelled cookstove intervention to prevent adoption of efficient Cookstoves? Evidence from a randomized controlled pneumonia in children under 5 years old in rural Malawi [the cooking and trial in Uganda. J Health Commun. 2015;20(sup1):55–66. pneumonia study]: a cluster randomised controlled trial. Lancet. 2017; 32. Bensch G, Peters J. A Recipe for Success? Randomized Free Distribution of 389(10065):167–75. Improved Cooking Stoves in Senegal, Ruhr Economic Papers, No. 325, ISBN 10. Smith KR, McCracken JP, Weber MW, Hubbard A, Jenny A, Thompson LM, et 978-3-86788-374-0, RWI, Essen; 2012. http://dx.doi.org/10.4419/86788374. al. Effect of reduction in household air pollution on childhood pneumonia 33. Pine K, Edwards R, Masera O, Schilmann A, Marrón-Mares A, Riojas- in Guatemala [RESPIRE]: a randomised controlled trial. Lancet. 2011;378: Rodríguez H. Adoption and use of improved biomass stoves in Rural 1717–26. Mexico. Energy for sustainable development 2011;15(2):176–183. 11. Pope D, Bruce N, Dherani M, Jagoe K, Rehfuess E. Real-life effectiveness of 34. Gordon J, Hyman J. The stoves are also stacked: evaluating the energy ‘improved’stoves and clean fuels in reducing PM 2.5 and CO: systematic ladder, cookstove swap-out programs and social adoption preferences in review and meta-analysis. Environ Int. 2017;101:7–18. the cookstove literature. J Environ Invest. 2012;3(1):17. 12. Masera OR, Saatkamp BD, Kammen DM. From linear fuel switching to 35. Beyond Distribution: Ensuring and Evaluating the Adoption of Clean multiple cooking strategies: a critique and alternative to the energy ladder Cooking and Its Benefits [Internet]. Lima, Peru: Global Alliance for Clean model. World Dev. 2000;28:2083–103. Cookstoves; 2015. Available from: https://cleancookstoves.org/binary-data/ 13. Ruiz-Mercado I, Masera O, Zamora H, Smith KR. Adoption and sustained use RESOURCE/file/000/000/379-1.pdf. Accessed 9 July 2015. of improved cookstoves. Energy Policy. 2011;39(12):7557–66. 36. International Energy Agency, Organisation for Economic Co-operation and 14. Ruiz-Mercado I, Masera O. Patterns of stove use in the context of fuel– Development World energy outlook 2016. Paris: OECD/IEA; 2016. device stacking: rationale and implications. EcoHealth. 2015;12(1):42–56. 37. Mosler H-J. A systematic approach to behavior change interventions for the 15. Piedrahita R, Dickinson KL, Kanyomse E, Coffey E, Alirigia R, Hagar Y, et al. water and sanitation sector in developing countries: a conceptual model, a Assessment of cookstove stacking in northern Ghana using surveys and review, and a guideline. Int J Environ Health Res. 2012;22:431–49. stove use monitors. Energy Sustain Dev. 2016;34:67–76. 38. Jack DW, Asante KP, Wylie BJ, Chillrud SN, Whyatt RM, Quinn AK, et al. 16. Johnson MA, Chiang RA. Quantitative stove use and ventilation guidance Ghana randomized air pollution and health study [GRAPHS]: study protocol for behavior change strategies. J Health Commun. 2015;20(Suppl 1):6–9. for a randomized controlled trial. Trials. 2015;16(1) 17. Zhou Z, Dionisio KL, Arku RE, Quaye A, Hughes AF, Vallarino J, et al. 39. Ghana Statistical Service. Population and Housing Census: Final Results Household and community poverty, biomass use, and air pollution in Accra, [Internet]; 2010. Available from: http://www.statsghana.gov.gh/docfiles/ Ghana. Proc Natl Acad Sci. 2011;108(27):11028–33. 2010phc/2010_POPULATION_AND_HOUSING_CENSUS_FINAL_RESULTS.pdf . 18. Zhou Z, Dionisio KL, Verissimo TG, Kerr AS, Coull B, Howie S, et al. Chemical Accessed 13 Nov 2017. characterization and source apportionment of household fine particulate 40. GhanaWeb. Ghana: Geography, Location, weather etc [Internet]. Available matter in rural, peri-urban, and urban West Africa. Environ Sci Technol. 2014; from: http://www.ghanaweb.com/GhanaHomePage/geography/climate. 48(2):1343–51. php . Accessed 12 June 2017. 19. Puzzolo E, Stanistreet D, Pope D, Bruce N, Rehfuess E. Factors influencing 41. Van Vliet ED, Asante K, Jack DW, Kinney PL, Whyatt RM, Chillrud SN, et al. the large-scale uptake by households of cleaner and more efficient Personal exposures to fine particulate matter and black carbon in household energy technologies [Internet]. London: EPPI-Centre, Social households cooking with biomass fuels in rural Ghana. Environ Res. 2013; Science Research Unit, Institute of Education, University of London; 2013. 127:40–8. Available from: https://cleancookstoves.org/binary-data/RESOURCE/file/000/ 42. Carrión D, Asante KP, Dwomah R, Mujtaba M, Taiwah T, Jack DW. Predictors 000/319-1.pdf. Accessed 12 Nov 2013. of clean Cookstove use in a rural Ghanaian cohort. Manuscript in 20. Rehfuess EA, Puzzolo E, Stanistreet D, Pope D, Bruce NG. Enablers and preparation; 2018. barriers to large-scale uptake of improved solid fuel stoves: a systematic 43. Scherbaum CA, Ferreter JM. Estimating statistical power and required review. Environ Health Perspect. 2014 Feb;122(2):120–30. sample sizes for organizational research using multilevel modeling. Organ 21. Lewis JJ, Pattanayak SK. Who adopts improved fuels and cookstoves? A Res Methods. 2009;12:347–67. systematic review. Environ Health Perspect. 2012;120(5):637. 44. Mosler HJ, Contzen N. Systematic behavior change in water, sanitation and 22. Jeuland M, Pattanayak SK, Bluffstone R. The economics of household air hygiene. A practical guide using the RANAS approach. ESI 3.1: Catalog of pollution. Annu Rev Resour Econ. 2015;7(1):81–108. behavior change techniques (BCTs) [Internet]. Dübendorf, Switzerland: 23. Kelly MP, Barker M. Why is changing health-related behaviour so difficult? Eawag; 2016. Available from: http://www.eawag.ch/fileadmin/Domain1/ Public Health. 2016;136:109–16. Abteilungen/ess/schwerpunkte/ehpsy/publ/Guideline/ESI_3.1_Catalog_of_ 24. Petty RE, Brinol P. Attitude change. Adv Soc Psychol State Sci. 2010:217–59. behavior_change_techniques__BCTs_.pdf. Accessed 13 June 2017. Carrión et al. BMC Public Health (2018) 18:689 Page 11 of 11 45. Mosler HJ, Contzen N. Systematic behavior change in water, sanitation and hygiene. A practical guide using the RANAS approach [Internet]. Dübendorf, Switzerland: Eawag; 2016. Available from: http://www.eawag.ch/fileadmin/ Domain1/Abteilungen/ess/schwerpunkte/ehpsy/publ/Guideline/Guideline_ V1.0_final_25_08_2016_druck_3.pdf.. Accessed 13 June 2017. 46. Sovacool BK, Drupady IM, ebrary Inc. Energy access, poverty, and development the governance of small-scale renewable energy in developing Asia. Surrey: Ashgate; 2012. xxii, 306 p. [Ashgate studies in environmental policy and practice] 47. Otoo M. NPA Launches National Consumer Week Celebration [Internet]. 10 November 2014, Modern Ghana. Available from: https://www.modernghana. com/news/580132/1/npa-launches-national-consumer-week.html. Accessed 13 June 2017. 48. Simon GL, Bailis R, Baumgartner J, Hyman J, Laurent A. Current debates and future research needs in the clean cookstove sector. Energy Sustain Dev. 2014;20:49–57. 49. Smith KR, Dutta K. Cooking with gas. Energy Sustain Dev. 2011 Jun 1;15(2): 115–6. 50. Hollada J, Williams NK, Miele HC, Danz D, Harvey AS, Checkley W. Perceptions of improved biomass and liquefied petroleum gas stoves in Puno, Peru: implications for promoting sustained and exclusive adoption of clean cooking technologies. Int J Environ Res Public Health 2017;14. 51. World LPG Association. Accelerating the LPG Transition: Global Lessons from Innovative Business and Distribution Models [Internet]; 2015. Available from: https://www.wlpga.org/wp-content/uploads/2015/09/accelerating-the-lpg- transition-2015-light.pdf . Accessed 13 June 2017. 52. ENERGIA International. Reaching the last mile in the extremely energy poor Asia-Pacific [Internet]; 2017. Available from: http://www.energia.org/ reaching-last-mile-extremely-energy-poor-asia-pacific/ . Accessed 13 June 53. Accenture Development Partnerships. Global Alliance for Clean Cookstoves: Brazil Feasibility Study [Internet]; 2011. Available from: https:// cleancookstoves.org/binary-data/RESOURCE/file/000/000/156-1.pdf . Accessed 13 June 2017. 54. Iacus SM, King G, Porro G, Katz JN. Causal inference without balance checking: coarsened exact matching. Polit Anal. 2012:1–24. 55. Ivers NM, Halperin IJ, Barnsley J, Grimshaw JM, Shah BR, Tu K, et al. Allocation techniques for balance at baseline in cluster randomized trials: a methodological review. Trials. 2012;13:120. 56. Moulton LH. Covariate-based constrained randomization of group- randomized trials. Clin Trials. 2004;1(3):297–305. 57. Lorenz E, Gabrysch S. Covariate-constrained randomization routine for achieving baseline balance in cluster-randomized trials. Stata Journal. 2017; 17(2):503–10. 58. Miller G, Mobarak AM. Gender differences in preferences, intra-household externalities, and low demand for improved cookstoves. National Bureau of Economic Research; 2013. Available from: http://www.nber.org/papers/ w18964 . Accessed 11 Apr 2016. 59. Pachauri S, Rao ND. Gender impacts and determinants of energy poverty: are we asking the right questions? Curr Opin Environ Sustain. 2013;5(2): 205–15. 60. Doss C. Intrahousehold bargaining and resource allocation in developing countries. World Bank Res Obs. 2013;28(1):52–78. 61. Tolhurst R, Amekudzi YP, Nyonator FK, Squire SB, Theobald S. “He will ask why the child gets sick so often”: the gendered dynamics of intra- household bargaining over healthcare for children with fever in the Volta region of Ghana. Soc Sci Med. 2008;66(5):1106–17. 62. Mani A. Mine, yours or ours? The efficiency of household investment decisions: an experimental approach. 2011; 63. Hoel JB. Heterogeneous households: a within-subject test of asymmetric information between spouses in Kenya. J Econ Behav Organ. 2015;118: 123–35. 64. Ruiz-Mercado I, Canuz E, Smith KR. Temperature dataloggers as stove use monitors [SUMs]: field methods and signal analysis. Biomass Bioenergy. 2012;47:459–68. 65. Rosenbaum J, Derby E, Dutta K. Understanding consumer preference and willingness to pay for improved cookstoves in Bangladesh. J Health Commun. 2015;20(sup1):20–7. 66. Ezzati M, Baumgartner JC. Household energy and health: where next for research and practice?. The Lancet. 2017;389(10065):130–2.
BMC Public Health – Springer Journals
Published: Jun 4, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera