Science of the Total Environment 716 (2020) 135338

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Knowledge and behavioural interventions to reduce human health risk

from private groundwater systems: A global review and pooled analysis
based on development status
S. Mooney a, C.P. McDowell b, J. O’Dwyer c,d,e, P.D. Hynds a,e,⇑
a
Environmental Sustainability & Health Institute, Technological University Dublin, Dublin, Ireland
b
School of Architecture, Planning and Environmental Policy, University College Dublin, Dublin, Ireland
c
School of Biological, Earth and Environmental Sciences, University College Cork, Cork, Ireland
d
Environmental Research Institute, University of Cork, Cork, Ireland
e
Irish Centre for Research in Applied Geosciences (iCRAG), University College Dublin, Dublin, Ireland

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 PICO review identified 40 distinct

private groundwater interventions.
 Just one identified intervention
associated with microbial
contamination of private
groundwater.
 Mean behavioural and knowledge
attainment across interventions was
53% and 48%, respectively.
 Interventions in economically
developed regions exhibited higher
behavioural outcomes.
 Practical interventions allied with
both large- and local-scale
awareness-raising campaigns
represent an optimum approach.

a r t i c l e i n f o a b s t r a c t

Article history: Groundwater contamination constitutes a significant health risk for private well users residing in rural
Received 29 August 2019 areas. As the responsibility to safeguard rural private domestic groundwater typically rests with non-
Received in revised form 31 October 2019 expert homeowners, interventions promoting risk mitigation and awareness represent the most viable
Accepted 31 October 2019
means of preventing supply contamination. However, no global review or pooled analyses of these inter-
Available online 23 November 2019
ventions has been undertaken to date. The current study sought to identify and quantify the performance
of private well interventions from 1990 to 2018 via a global systematised review and pooled analysis. The
PICO (Population-Intervention-Comparison-Outcome) approach was employed for literature identifica-
tion. Relevant studies were statistically analysed across two quantitative outcome (performance) types,
namely knowledge and behaviour, controlling for intervention characteristics and country development
status. Mean behavioural and knowledge attainment across interventions was 53% and 48%, respectively,
with interventions in economically developed regions exhibiting higher behavioural outcomes (56% vs.
45%) than those in developing regions. Geographically, interventions were located in southern or south-
east Asia (n = 23), North America (n = 15), Central America (n = 1) and Africa (n = 1), with none identified
in Australia/Oceania, Europe, or South America. Behavioural outcomes were significantly associated with
presence of educational/research coordinator (p = 0.023), with these interventions attaining higher levels
of efficacy (+74%) than those implemented by other coordinator types. Findings indicate that instructor-

⇑ Corresponding author.
E-mail address: [email protected] (P.D. Hynds).

https://doi.org/10.1016/j.scitotenv.2019.135338
0048-9697/Ó 2019 Elsevier B.V. All rights reserved.
2 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338

led, practical interventions allied with both large- and local-scale awareness-raising campaigns represent
an optimum approach for future private well risk interventions. Subsequent adoption of such interven-
tions may lead to increased levels of private well maintenance and provide a point of reference for myriad
water and health communication contexts.
Ó 2019 Elsevier B.V. All rights reserved.

1. Introduction associated with their own suite of potential hazards to private

groundwater; for example, DWWTS-originated contamination
Groundwater is the most extracted raw material on earth and stemming from rapid exurban housing development and emerging
represents the largest global reservoir of accessible freshwater organic contaminants (EOCs), e.g. industrial compounds, and
for human consumption, currently serving an estimated 2.2 billion intensive agriculture (Lapworth et al., 2012). In order to effectively
people (Gleeson et al., 2016; Murphy et al., 2017). With a signifi- minimise groundwater contamination and consequent health
cant portion of historical groundwater abstraction occurring in impacts on a global scale, it is important to acknowledge both
the last 50 years due to rising populations and dwindling surface regional commonalities and variations.
water resources, established reserves have become overexploited Due to the absence of private groundwater regulation in many
(Gorelick & Zheng, 2015). As a result, groundwater quality has territories, the onus to protect private rural groundwater largely
become severely compromised in many regions, with improved lies with non-expert owners and users. Interventions entailing
supply management now regarded as a key environmental health public engagement (e.g. information campaigns, well testing ser-
challenge for the 21st century (Gorelick & Zheng, 2015; Jakeman vices) to promote protective behaviours and risk awareness may
et al., 2016). Much of the recent research pertaining to global therefore constitute the most feasible approach to ensuring
groundwater management has focused on predicting and mod- improved rural groundwater quality – irrespective of region of
elling aquifer withdrawals and developing options for recovery implementation (Kreutzwiser et al., 2011; Mitchell et al., 2012).
and restoration (Niu et al., 2014). Of more immediate concern is The necessity for these initiatives has been highlighted in several
the widespread susceptibility of potable sources to contamination previous studies reporting pronounced shortcomings in risk
posing human health risks, particularly in light of shifting land-use awareness and protective actions, with many recommending
patterns and climate change (Green et al., 2011; O’Dwyer et al., renewed or novel outreach strategies (Charrois, 2010; Malecki
2016; Andrade et al. 2018). et al., 2017). However, in spite of the growing acknowledgement
Unprecedented rates of groundwater abstraction in areas char- that enhanced engagement with private well users is necessary,
acterised by high geochemical concentrations, expanding agro- benchmarks for effective private groundwater interventions (lead-
industrial activities and climate change-induced changes in precip- ing to ‘‘bottom-up” mitigation) remain elusive and existing frame-
itation patterns have resulted in the wide proliferation of chemical works inadequate (Mitchell et al., 2012; Hynds et al., 2018).
and microbial contaminants in groundwater supplies Quantification of current success rates and outcome predictors
(Schwarzenbach et al., 2010; Gorelick & Zheng, 2015). An increas- associated with rural groundwater interventions may represent a
ing number of well users worldwide have consequently suffered benchmark for future hydrogeological investigations and manage-
severe health impacts due to chronic exposure to naturally occur- ment strategies addressing the intersect between groundwater and
ring chemicals such as arsenic and nitrate and contraction of end-users. While previous studies have identified contextual fac-
waterborne diseases (Jakeman et al., 2016). Global estimates sug- tors surrounding well user outreach (e.g. Morris et al., 2016), to
gest that over 150 million people currently ingest arsenic from date, no broad empiric analysis of rural groundwater interventions
groundwater drinking supplies exceeding permissible limits set and associated behavioural and knowledge outcomes exists in the
by the WHO (Ravenscroft et al., 2009). Moreover, a recent global literature. The current study responds to this deficit via a global
review of groundwater-related enteric disease outbreaks identified systematised review and pooled-analysis to establish the perfor-
649 events within the published literature during the period 1948- mance of private well user interventions designed to engender
2015 and a steadily increasing incidence of groundwater-related increased source protection and risk awareness.
acute gastrointestinal infections (AGI) (Murphy et al., 2017). Rural
populations relying on private, unregulated wells are particularly
2. Methods
susceptible to illness due to the wide range of proximate contam-
inant sources (e.g. domestic wastewater treatment systems
A systematised scoping review protocol was adapted and devel-
(DWWTSs), agriculture) and geographical/infrastructural obstacles
oped from previous studies (Elamin et al., 2009; Hynds et al.,
(e.g. isolation from major population centres and services) hinder-
2014). The review protocol was accepted and registered on PROS-
ing ‘‘top-down” groundwater management and policy implemen-
PERO (Registration Number CRD42017078019), and comprised
tation (Pearson et al., 2011; Valle Junior et al., 2014).
four distinct process steps, namely: Literature Identification (Sec-
Accordingly, private groundwater interventions (i.e. initiatives to
tion 2.1), Study Selection (Section 2.2), Data Extraction (Section 2.3)
promote source safety and protection) are being increasingly
and Pooled Analysis (Section 2.4).
implemented in rural areas to compensate for the widespread
absence of regulatory tools (Morris et al., 2016).
While many groundwater management challenges are fre- 2.1. Literature identification
quently associated with socioeconomically ‘‘developing” regions,
concomitant concerns are also reported in the more affluent ‘‘de- The first stage of an evidence-based review process comprises
veloped” nations (Shankar et al., 2014; WHO/UNICEF, 2015). Addi- development of an answerable question to facilitate identification
tionally, some of the most prevalent groundwater contaminants of relevant literature (Moher et al., 2009). The present review was
(e.g. naturally occurring arsenic) encountered in developing coun- guided by the following research question:
tries (e.g. Bangladesh) also occur in high concentrations in devel- What interventions have been employed to improve awareness
oped countries (e.g. United States) (Shankar et al., 2014; Henry & and protective behaviour(s) among rural private groundwater
Suk, 2018). Conversely, rural regions in developed countries are users and what levels of success have been achieved?
 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338 3

Due to the significant public health dimension of private Table 1

groundwater interventions, the research question and literature Literature review search terms and correspondent PICO classification and description.

identification process employed the PICO (Population, Interven- Primary Term Description Search Terms
tion, Comparison, Outcomes) approach (Schardt et al., 2007; Term
Higgins & Green, 2011). The PICO method and its application Population Rural private groundwater Borehole users; groundwater
within the current study, including explicit search terms and asso- supply (borehole, spring or supply; private groundwater;
ciated PICO components, are presented in Figure 1 and Table 1. In well) users private well owners; spring
water; tube wells
light of the multidisciplinary and global span of the review and Intervention National and subnational Education; engagement;
consequent requirement for wide journal coverage, potentially rel- outreach strategies information campaign;
evant literature was searched using the Google Scholar search targeting private well users intervention; outreach; risk
engine. While the authors acknowledge the weaknesses of Google (information campaigns, communication
workshops etc.)
Scholar as a systematic search tool (e.g. limited search algorithms
Comparison Differences between Contamination; inspection;
and filters), assuring a blanket aggregation of studies was consid- interventions in type, registration; sampling;
ered optimal for maximum article retrieval and deemed to negate region, reach etc. testing; treatment
these limitations (Haddaway et al., 2015). Databases characterised Outcomes Impact on well user Awareness; behaviour;
by a comparatively specialised, restricted focus (e.g. Scopus, behaviour and knowledge knowledge; risk prevention;
risk reduction
PubMed) were hence not consulted (Falagas et al., 2008). A review
period of 29 years (1990-2018) was used to ensure comprehensive
capture of temporally relevant data. Search combinations incorpo-
Table 2
rated a minimum of four individual search terms (one per PICO Eligibility (inclusion/exclusion) criteria employed for literature screening.
component) (Table 1).
Explicit eligibility (inclusion/exclusion) criteria were used for Inclusion criteria Exclusion criteria

article screening (Table 2). Only peer-reviewed academic journal Literature Peer-reviewed academic Grey literature
articles written in English evaluating measurable interventions journal articles
Language English Non-English
intended for private groundwater supply users during or after
Population Private groundwater supply Surface water and public
1990 were considered. Grey literature (e.g. technical reports, work- users residing in rural areas groundwater users +
ing papers, conference papers/abstracts, etc.) was excluded to households residing in urban
avoid publication bias, skewed intervention coverage, and/or inad- or peri-urban areas
Evaluation Quantitative investigations of Investigations with insufficient
equate study design. Investigations located in non-rural (i.e. urban
interventions intended to or non-quantitative data
or peri-urban) areas and/or failing to present demarcated quantita- promote groundwater supply relating to intervention
tive data specific to private well users were not considered appro- user stewardship outcomes
priate for inclusion. Likewise, interventions not specifically Period 1990-present Pre-1990 (intervention and
directed at private groundwater sources and/or users were article)

excluded.

ing omission of grey literature (-90 studies) and addition of

2.2. Study selection (screening) relevant journal article abstracts/titles from study reference lists
(+81 studies), 270 articles remained for the second phase of litera-
The literature search (SM, CM) returned 15,946 potentially rel- ture screening (full article analysis).
evant studies, of which 4,080 were retained for screening based on Full articles were analysed using established eligibility criteria,
eligibility (Table 2). The first stage of literature screening was with studies independently assessed by three authors (SM, CM, PH)
based on an independent assessment of article titles and abstracts and, where divergence arose, iteratively deliberated until consen-
(Figure 2), with 279 abstracts deemed to fulfil inclusion criteria. sus was reached. The final stage of screening was predicated on
‘‘Legacy” searches were subsequently undertaken to ensure identi- suitability of interventions and evaluations, resulting in the inclu-
fication of studies undetected during electronic searches. Follow- sion of 31 articles containing 40 distinct interventions.

Figure 1. The PICO approach as applied for literature identification and data extraction in the current study.
4 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338

Figure 2. Schematic of study selection protocol.

2.3. Data extraction differ in orientation and effectiveness (Oxman et al., 2009;
Waddington et al., 2012).
Overall, 74 variables were extracted under nine primary cate- Intervention reach (i.e. geographic extent of influence) was cat-
gories (Appendix). Interventions were coded with respect to coun- egorised with reference to the WHO’s ExpandNet Framework for
try development status following the emergence of a natural intervention scaling and divided into three classifications: large
dichotomy between high and low/medium development nations. (national/state/regional), medium (sub-regional/county) and small
The separation was deemed to increase accessibility for prospec- (community/local) (WHO, 2009). Intervention mechanisms (i.e.
tive private groundwater intervention coordinators, as interven- constituent activities of overarching interventions) were cate-
tions in countries of contrasting socio-economic interest, gorised based on public engagement typologies developed by
notwithstanding potential similarities/transferability, are likely to Rowe & Frewer (2005), with two mechanism types (electronic
 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338 5

and print) based upon uni-directional information flow from inter- ope, South America or the Australian sub-continent were identified
vention coordinator to audience (recipient population) and one (Figure 3). Chemical and microbial contaminants were reported in
mechanism type (interpersonal) based upon reciprocal (bi- 83% (n = 33) and 3% (n = 1) of interventions, respectively, with
directional) information flow. A fourth mechanism type (conve- remaining studies failing to specify contaminant type. Drilled bore-
nience) was designated for intervention activities entailing direct holes comprised the majority (80%, n = 32) of private groundwater
support (e.g. provision of contaminant treatment system) and/or supplies for which interventions were reported. Well depths and
practical action (e.g. well installation) on behalf of well users. Clas- local/regional (hydro)geological information was reported for just
sification of intervention stakeholders, i.e. coordinators (bodies 18% (n = 7) and 13% (n = 5) of interventions, respectively. Six inter-
responsible for implementing interventions) broadly emulated ventions were undertaken in areas currently subject to private
the stakeholder classes set out by Bunn et al. (2002), as follows: groundwater regulation (United States), including mandatory well
community body, educational/research body, government body, non- water testing transactions (10%, n = 4) and well construction stan-
profit body and other body. Extracted outcomes comprised mea- dards (5%, n = 2).
sured behaviours and knowledge levels explicitly related to well Seven intervention types (i.e. distinct outreach strategies) were
safety, maintenance and health risks and were selected based on identified, with information campaigns (65%, n = 26) and well test-
their assigned importance to intervention objectives by study ing services (65%, n = 26) the most frequently implemented
authors. Where primary behavioural and knowledge outcomes (Table 4). With respect to intervention reach, interventions were
were not specified, multiple measures were combined and aver- most often oriented at the subregional/county (medium) scale
aged for each respective outcome type. (43%, n = 17). Of the four intervention mechanism types (i.e. chan-
Dummy variables were developed for seven data categories; in nels of engagement), the most commonly employed was conve-
instances where relevant information (e.g. regulatory status of pri- nience (83%, n = 33), followed by interpersonal (60%, n = 24),
vate groundwater supplies in intervention area) was not provided print (45%, n = 18) and electronic (25%, n = 10). Interventions were
in articles but available elsewhere, relevant sources (e.g. UNDP incentivised in 85% (n = 34) of cases, with incentives ranging from
Human Development Index) were consulted (UNDP, 2016). For practical (e.g. provision of water sample bottle) to financial (e.g.
numerical (continuous) variables (e.g. intervention duration), subsidised water treatment), and primarily comprising direct sup-
intervals and units of measurement were discretised into appropri- plementary measures (e.g. provision of water sampling kits) to
ate ranges. Upon completion of data extraction and division of alleviate well maintenance costs. Pro bono incentives (e.g. free well
interventions by country development status, 69 variables were testing) were reported in 10% (n = 4) of cases. With respect to inter-
retained for coding and analyses, with five variables (hydrogeolog- vention stakeholders (see Section 2.3), government bodies were
ical setting, intervention cost, presence of disease/illness, number the most frequently referenced coordinator type (48%, n = 19), fol-
of wells, well depth) excluded due to incomplete records (available lowed by educational/research (45%, n = 18) and non-profit (38%, n
in <50% of studies). = 15). Community actors (i.e. volunteer groups involved in inter-
ventions downstream) were noted in 20% (n = 8) of interventions,
all of which were implemented in developing countries.
2.4. Pooled Analysis

Data were numerically coded and imported to IBM SPSS Statis-

3.2. Development status
tics 26 (Armonk, NY: IBM Corp) for statistical analyses. Descriptive
statistical functions (i.e. box and whisker plots displaying the med-
A total of 15 identified interventions (38%) originated from
ian, interquartile ranges and present extreme/outlier values and
developed (high development) countries and 25 (63%) from devel-
stem-and-leaf plots) were employed to detect outliers across
oping (low/medium development) countries (Table 4). More than
results per outcome type (behaviour and knowledge). Upon
three quarters of information campaigns (77%, n = 20) took place
removal of outliers, outcomes were analysed for normality through
in developing countries, while all workshops (n = 4) were under-
visual inspection of Q-Q plots and the Shapiro-Wilk test. As overall
taken in developed countries. A difference approaching statistical
outcomes followed a normal distribution, parametric statistical
significance was found between primary (i.e. overarching) inter-
testing was employed.
vention type (v2 = 11.006, p = 0.051) and development status
The Pearson chi-square test of independence was used to test
(Table 6); source installation, source remediation and source
for associations between dichotomous (e.g. developed/developing)
treatment-based initiatives were exclusive to developing countries
and nominal (e.g. intervention type) categorical variables. One-
while all interventions entailing workshops as a strategy (primary
way analysis of variance (ANOVA) tests were used to explore rela-
or otherwise) were documented in developed countries. Over three
tionships between intervention characteristics (categorical vari-
quarters (n = 19) of interventions in developing countries com-
ables) and outcomes (continuous variables). One-way ANCOVAs
prised >1 intervention type (i.e. multimodal strategies) compared
examined adjusted associations (study design and country devel-
to 27% (n = 4) in developed countries (v2 = 9.337, p = 0.002). Inter-
opment status) between intervention characteristics and knowl-
vention reach was also significantly associated with development
edge and behaviour outcomes. Study design denotes presence or
category for large-scale interventions (v2 = 6.009, p = 0.014), with
absence of baseline data in intervention evaluations. The level of
47% (n = 7) of interventions undertaken at this scale in developed
significance was set at p <0.05 by convention.
countries compared to 12% (n = 3) in developing countries.
Convenience mechanisms were encountered in all interventions
3. Results in developing countries (v2 = 14.141, p = <0.001), with print mech-
anisms more prevalent in developed countries (87%, n = 13) (v2 =
3.1. Intervention characteristics 8.095, p = 0.004). Presence of an incentive was also significantly
associated with development status (v2 = 6.327, p = 0.012); almost
The review process yielded 40 distinct interventions (Figure 2), all interventions in developing countries (96%, n = 24) contained
of which 58% (n = 23) were undertaken in southern or south- 1 incentive compared to 67% (n = 10) in developed countries. In
eastern Asia, with remaining interventions from the United States terms of coordinator type, presence of a non-profit body (v2 =
(28%, n = 11), Canada (10%, n = 4), Guatemala (3%, n = 1) and 11.111, p = 0.001) was significantly associated with development
Uganda (3%, n = 1) (Table 3). Notably, no interventions from Eur- status. Non-profit bodies had a coordinating role in 60% (n = 15)
6 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338

Table 3
Characteristics of identified studies (n = 31) delineated by country development status.

Article No. of interventions Study location Contaminants Private groundwater

Country Type Origin Supply type Supplyregulation
Section 1: Developed (n = 12)
Benham et al. (2016) 1 US - - Drilled, dug, spring Yes
Clemens et al. (2007) 1 US - - Drilled No
Flanagan et al. (2015) 1 US Chemical Geogenic Drilled, dug No
Flanagan et al. (2016a) 1 US Chemical Geogenic - Yes
Flanagan et al. (2016b) 2 US Chemical Geogenic - Yes
He et al. (2018) 1 US Chemical Geogenic Drilled, dug Yes
Hexemer et al. (2008) 2 Canada Chemical Geogenic - No
McCann & Gold (2012) 1 US - - Drilled Yes
Paul et al. (2015) 1 US Chemical Geogenic Drilled, dug Yes
Renaud et al. (2010) 2 Canada Chemical Geogenic - No
Schubert et al. (1999) 1 US Chemical Anthropogenic Drilled No
Swistock et al. (2001) 1 US - - Drilled No
Section 2: Developing (n = 19)
Balasubramanya et al. (2014) 1 Bangladesh Chemical Geogenic Drilled No
Barnwal et al. (2017) 1 India Chemical Geogenic Drilled No
Bennear et al. (2013) 2 Bangladesh Chemical Geogenic Drilled No
Brown et al. (2009) 1 Cambodia - - Drilled, dug No
Caldwell et al. (2005) 1 Bangladesh Chemical Geogenic Drilled, dug No
Chen et al. (2007) 1 Bangladesh Chemical Geogenic Drilled No
George et al. (2012) 2 Bangladesh Chemical Geogenic Drilled No
George et al. (2013) 3 Bangladesh Chemical Geogenic Drilled No
Hadi (2003) 1 Bangladesh Chemical Geogenic Drilled No
Hoque et al. (2000) 1 Bangladesh Chemical Geogenic Drilled, dug No
Hoque et al. (2004) 1 Bangladesh Chemical Geogenic Drilled, dug No
Jakariya et al. (2007) 1 Bangladesh Chemical Geogenic Drilled, dug No
Johnston & Sarker (2007) 1 Bangladesh Chemical Geogenic Drilled, dug No
Jones & Khaira (2014) 3 Cambodia Chemical Geogenic Drilled No
Keskin et al. (2017) 1 Bangladesh Chemical Geogenic Dug, drilled No
Luby et al. (2008) 1 Guatemala - - Dug, spring No
Lule et al. (2005) 1 Uganda Microbial Anthropogenic Drilled, dug, spring No
Madajewicz et al. (2007) 1 Bangladesh Chemical Geogenic Drilled No
Mahmud et al. (2007) 1 Bangladesh Chemical Geogenic Drilled, dug No

Figure 3. Global distribution of identified private groundwater interventions (1990-2018).

 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338 7

Table 4
Intervention characteristics delineated by country development status.

Article Intervention General Constituent Stakeholders

* attributes activities
Reach Duration (months) Mechanism types # No. of incentives Coordinator body type(s) ǂ
Section 1: Developed
Benham et al. (2016) WTS, W Large 270 C, E, I, P 4 Edu/Res.
Clemens et al. (2007) W Large 18 E, I, P - Edu/Res., NPO
Flanagan et al. (2015) WTS Medium 60 C, P 2 Edu/Res., Govt.
Flanagan et al. (2016a) Reg. Large 141 P - Govt.
Flanagan et al. (2016b) IC, WTS a Small - C 1 Govt.
IC, WTS b Small - C 1 Govt.
He et al. (2018) WTS Large - C, E, P 1 Edu/Res.
Hexemer et al. (2008) WTS c Medium 0.23 C, I, P 5 Govt.
WTS d Medium 1 C, I, P 5 Govt.
McCann & Gold (2012) W Large 196 E, I, P 1 Edu/Res., Govt.
Paul et al. (2015) IC, WTS Small 19 C, I, P 4 Other
Renaud et al. (2010) IC Small 3 P 1 Govt.
IC Medium 1 E, P - Govt.
Schubert et al. (1999) IC Large 48 P - Govt.
Swistock et al. (2001) W Large 0.03 E, I, P - Edu/Res.
Section 2: Developing
Balasubramanya et al. (2014) WTS Medium 48 C 3 Govt.
Barnwal et al. (2017) IC, WTS Medium 42 C, I, P - Edu/Res.
Bennear et al. (2013) IC e Medium 2 C, I 1 Edu/Res.
IC f Medium 2 C, I 1 Edu/Res.
Brown et al. (2009) TSD Large 42 C 1 NPO
Caldwell et al. (2005) IC, WTS Large 30 C 1 Govt., NPO
Chen et al. (2007) IC, SI, WTS Medium 46 C, I 3 Edu/Res., Govt., NPO
George et al. (2012) IC, WTS g Medium 3 C, I 3 Edu/Res., NPO
IC, WTS h Medium 3 C, I 3 Edu/Res., NPO
George et al. (2013) IC i, WTS Small 3 C, I, P 2 Edu/Res., NPO
IC j, WTS Small 3 C, I, P 2 Edu/Res., NPO
WTS Small 3 C, P 2 Edu/Res., NPO
Hadi (2003) IC, SI, W Medium 12 C, I, P 1 Govt., NPO
Hoque et al. (2000) IC, SI, WTS Small 12 C, E, I 3 Edu/Res., Govt.
Hoque et al. (2004) IC, SI, W, WTS Small 13 C, I 4 NPO
Jakariya et al. (2007) IC, SI, WTS Medium 18 C, E, I 3 NPO
Johnston & Sarker (2007) IC, SI, WTS Medium 48 C, P 3 Govt.
Jones & Khaira (2014) IC, SI, WTS k Small - C, E, I, P 3 NPO
IC, WTS l Small - C, P 2 NPO
IC, WTS m Small - C, P 2 NPO
Keskin et al. (2017) IC, SI, WTS Large 84 C 3 Govt.
Luby et al. (2008) IC, TSD Medium 3 C, E, I, P 2 Govt., Other
Lule et al. (2005) IC, TSD Medium 20 C, I 2 Govt.
Madajewicz et al. (2007) IC, SI, WTS Small 12 C, I 3 Edu/Res., Govt., NPO
Mahmud et al. (2007) SR Medium 10 C, I 3 Edu/Res.

* IC = information campaign, Reg. = regulation, SI = source installation, SR = source remediation, TSD = treatment system distribution, W = workshop, WTS = well testing
service
# C = convenience, E = electronic, I = interpersonal, P = print
ǂ Comm. = community, Edu/Res. = educational/research, Govt. = government, NPO = non-profit organisation
Intervention differentiators (as defined by study authors):
a – high intensity well testing promotion, b – low intensity well testing promotion, c – pilot intervention, d – full intervention, e – households encouraged to seek water from
wells below national arsenic standard, f – households receive additional information relating to arsenic safety and encouraged to seek water from wells with lower levels of
arsenic (irrespective of national standard), g – performed by community members, h – performed by representatives from outside communities, i – household education, j –
household education and local media campaign, k – mitigation village, l – comparison village I, m – comparison village II

of interventions in developing countries but were noted for only interventions in developing countries exhibiting a mean of 44%
one intervention undertaken in a developed country. (SD = 28.30) and developed countries a mean of 47% (SD = 24.02)
(Table 7).
3.3. Intervention evaluation and efficacy
3.4. Behavioural outcomes
Behavioural outcomes were measured in 93% (n = 37) and
knowledge outcomes in 60% (n = 24) of interventions (Table 5). The mean level of behavioural change reported in developed
Outcomes were measured via knowledge, attitudes and/or prac- countries was 46% (SD = 22.49) compared to 56% (SD = 41.70) in
tices (KAP) surveying in 39 out of 40 intervention evaluations, with developing regions. Based on primary intervention type, highest
the remaining study (Paul et al., 2015) measuring behaviour mean behavioural outcomes were recorded for source remediation
change via number of well water samples submitted for testing. (135%, n = 1), followed by safe source installation (72%, SD = 48.85)
The most frequent behavioural outcome type was use of a safe and workshops (65%, SD = 16.38). In terms of reach, mean beha-
water source (n = 14), e.g. a newly abstracted well, while risk vioural outcomes were highest for small-scale interventions (64%,
awareness (n = 17) was the most common type of knowledge SD = 14.18), with large- and medium-reach interventions perform-
outcome. ing at 55% (SD = 32.79) and 43% (SD = 35.60), respectively.
Discounting outlier values (resulting in truncated mean values), Mean behavioural outcomes displayed a significant associated
the mean efficacy of all interventions was 45% (SD = 26.46), with with involvement of educational/research coordinators in inter-
8 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338

Table 5
Evaluation methods and outcomes (%) of identified studies delineated by country development status.

Article Intervention Evaluation Behaviour outcome Knowledge outcome

Method
Study design * Description % Description %
Section 1: Developed
Benham et al. (2016) WTS, W Survey Post 1 action taken # +70 - -
Clemens et al. (2007) W Survey Post 1 action taken # +82 1 action learned +92
Flanagan et al. (2015) WTS Survey Post 1 action taken # +73 Recall of well test result +67
Flanagan et al. (2016a) Reg. Survey Pre-post Well testing & treatment +16 Recall of well test result +1
Flanagan et al. (2016b) IC, WTS Survey Pre-post Well testing +32 Risk awareness +83
IC, WTS Survey Pre-post Well testing +16 Risk awareness +52
He et al. (2018) IC Survey Post Use of safe water source +29 - -
Hexemer et al. (2008) WTS Survey Post Well testing +42 - -
WTS Survey Post Well testing +47 - -
McCann & Gold (2012) W Survey Post Combined actions ǂ +43 - -
Paul et al. (2015) IC, WTS Water samples Pre-post Well testing +243 - -
Renaud et al. (2010) IC Survey Pre-post Well screening +300 Risk awareness 6
IC Survey Post - Risk awareness +6
Schubert et al. (1999) IC Survey Post 1 action taken # +30 Recall of well test result +66
Swistock et al. (2001) W Survey Post 1 action taken # +66 2 actions learned +71
Section 2: Developing
Balasubramanya et al. (2014) WTS Survey Post Use of safe water source +41 Recall of well test result +63
Barnwal et al. (2017) Survey Pre-post Use of safe water source +31 Recall of well test result +50
Bennear et al. (2013) IC Survey Pre-post Use of safe water source +29 - -
IC Survey Pre-post Use of safe water source +20 - -
Brown et al. (2009) TSD Survey Post Use of water treatment system +18 - -
Caldwell et al. (2005) IC, WTS Survey Pre-post Combined actions ǂ +113 Risk awareness +77
Chen et al. (2007) IC, SI, WTS Survey Pre-post Use of safe water source +58 - -
George et al. (2012) IC, WTS Survey Pre-post Use of safe water source +44 Risk awareness +68
IC, WTS Survey Pre-post Use of safe water source +63 Risk awareness +67
George et al. (2013) IC, WTS Survey Pre-post Purchase of well test +93 Risk awareness +41
IC, WTS Survey Pre-post Purchase of well test +93 Risk awareness +41
WTS Survey Pre-post Purchase of well test +53 Risk awareness +36
Hadi (2003) IC, SI, W Survey Pre-post - Risk awareness +306
Hoque et al. (2000) IC, SI, WTS Survey Pre-post Use of safe water source +144 - -
Hoque et al. (2004) IC, SI, W, WTS Survey Pre-post Use of safe water source +36 Risk awareness +78
Jakariya et al. (2007) IC, SI, WTS Survey Post Use of safe water source +50 - -
Johnston & Sarker (2007) IC, SI, WTS Survey Pre-post Use of safe water source +38 Risk awareness +71
Jones & Khaira (2014) IC, SI, WTS Survey Post Use of safe water source +90 Risk awareness +42
IC, WTS Survey Post Use of safe water source +25 Risk awareness +5
IC, WTS Survey Post - Risk awareness +30
Keskin et al. (2017) IC, SI, WTS Survey Pre-post Breastfeeding duration +85 - -
Luby et al. (2008) IC, TSD Survey Post Use of treatment system +3 - -
Lule et al. (2005) IC, TSD Survey Pre-post Combined actions ǂ 29 - -
Madajewicz et al. (2007) IC, SI, WTS Survey Pre-post Use of safe water source +60 Risk awareness 6
Mahmud et al. (2007) SR Survey Pre-post Combined actions ǂ +135 - -

IC = information campaign, Reg. = regulation, SI = source installation, SR = source remediation, TSD = treatment system distribution, W = workshop, WTS = well testing service.
* Pre and post indicates employment of baseline data.
# Well owners demonstrating adoption of at least one well maintenance behaviour (e.g. periodic testing, supply inspection).
ǂ Mean percentage of supply maintenance following amalgamation of percentages for multiple actions.

ventions (F = 5.712, p = 0.023). Interventions under sole or co- electronic-based mechanisms (64%, SD = 40.35), followed by inter-
supervision of educational/research coordinators attained a mean personal (58%, SD = 39.38), convenience (53%, SD = 37.93) and print
behavioural increase of 66% (SD = 34.29) compared to 38% (SD = (51%, SD = 27.59) (Fig 5).
38.67) for those comprising other coordinators. Behavioural out-
comes were also significantly associated with involvement of an
4. Knowledge outcomes
educational/research coordinator when controlling for study
design (i.e. presence of baseline or pre-intervention data) and devel-
The mean knowledge outcome reported in developed countries
opment status (Table 8). Interventions involving a non-profit
was 48% (SD = 37.55) compared to 47% (SD = 25.75) in developing
organisation in implementation accomplished a mean behavioural
regions. Highest mean knowledge outcomes were attained by
outcome of 63% (SD = 28.08) while those comprising other coordi-
interventions comprising workshops (82%, SD = 14.85), signifi-
nator types attained a mean behavioural outcome of 46% (SD =
cantly outperforming information campaigns (41%, SD = 28.23)
40.044). Interventions involving a government coordinator accom-
and well testing services (47%, SD = 28.88). Large-scale interven-
plished a mean behavioural increase of 48% (SD = 40.68) in contrast
tions were associated with the greatest increase in knowledge
to 57% (SD = 32.09) for interventions implemented by different
(61%, SD = 35.15), followed by medium (56%, SD = 23.08) and
coordinator types (Table 9).
small-scale initiatives (36%, SD = 29.73).
Interventions employing 1 incentive-based measure (e.g. free
A mean knowledge outcome of 53% (SD = 26.93) was attained
well testing) accomplished a mean behavioural change of 54% (SD
by interventions under supervision of an educational/research
= 37.75) compared to 45% (SD = 27.713) for interventions foregoing
coordinator compared to 44% (SD = 32.81) for those involving other
incentives. With respect to intervention mechanisms (Figure 4),
coordinator types. Interventions under supervision of a govern-
highest rates of behaviour were associated with presence of
mental organisation in accomplished a mean knowledge outcome
 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338 9

Table 6
Bivariate analysis of associations between intervention characteristics and development type.

Variable(s) Development category1 Test statistic (v2) P value

Developed (%) Developing (%)
Primary intervention type 11.006 0.051
Information campaign 6 (40) 10 (40) 0 1
Regulation 1 (7) – 1.709 0.191
Source installation – 4 (16) 2.667 0.102
Source remediation – 1 (4) 0.615 0.433
Treatment system distribution – 3 (12) 1.946 0.163
Well testing service 4 (27) 7 (28) 0.008 0.927
Workshop 4 (27) – 7.407 0.006*
>1 intervention type 4 (27) 19 (76) 9.337 0.002*
Intervention reach 6.174 0.046*
Large 7 (47) 3 (12) 6.009 0.014*
Medium 4 (27) 13 (52) 2.462 0.117
Small 4 (27) 9 (36) 0.372 0.542
Convenience mechanism 8 (53) 25 (100) 14.141 0.000*
Electronic mechanism 6 (40) 4 (16) 2.88 0.09
Interpersonal mechanism 8 (53) 17 (68) 1.778 0.182
Print mechanism 13 (87) 10 (40) 8.335 0.004*
Incentive 10 (67) 24 (96) 6.327 0.012*
Edu/res. coordinator 6 (40) 12 (48) 0.242 0.622
Govt. body coordinator 10 (67) 10 (40) 2.667 0.102
NPO coordinator 1 (7) 15 (60) 11.111 0.001*
1
Frequency of stated variables by development status.
*
Significant at <0.05 level.

Table 7
Mean and truncated mean intervention outcome percentage and bivariate analysis by country development status (t-tests employed to identify the difference between truncated
mean outcomes based on country development status).
1 1
Outcome (%) DevelopmentStatus N TruncatedMean TruncatedStd. Dev. Test statistic (t) P value
Behaviour Developed 14 45.50 22.49
Developing 23 56.22 41.70 0.826 0.415
All 37 52.54 36.27
Knowledge Developed 9 48.00 37.55
Developing 15 47.36 25.75 0.049 0.961
All 24 47.61 30.80
Overall Developed 15 47.08 24.02
Developing 25 44.18 28.30 0.309 0.760
All 40 45.26 26.46
1
Outliers were identified via box and whisker plots and stem-and-leaf plots in SPSS and removed prior to analysis.

of 43% (SD = 36.14) while those comprising other coordinator types lights a series of unique characterisations and limitations in both
attained a mean behavioural outcome of 52% (SD = 21.14). intervention reporting and scope which also merit reflection.
No difference in mean knowledge attainment was identified
based on presence of absence of incentives, with a mean perfor- 5.1. Global distribution of interventions
mance of 48% recorded interventions with and without incentives
(SD = 28.66; SD = 36.78). Interpersonal mechanisms comprised the The continents of North America, Asia and Africa collectively
most successful engagement channel in knowledge attainment constitute approximately 80-90% of the estimated 2.2 billion
(54%, SD = 27.43), followed by electronic (53%, SD = 37.30), conve- groundwater users worldwide (Murphy et al., 2017). However,
nience (51%, SD = 25.11) and print based mechanisms. (41%, SD = the interventions identified in this review were largely concen-
29.59). trated within the Indian subcontinent and the North-eastern Uni-
ted States, thus accounting for a relatively limited proportion of
5. Discussion global private groundwater users. The absence of studies evaluat-
ing interventions in Oceania, Europe and South America, may be
The current study is the first to quantify the efficacy of private due to growing urbanisation, lower private groundwater reliance,
well interventions for improving awareness and protective beha- absence of research-led interventions, publication bias, and war-
viours among private groundwater users. Interventions imple- rant future research attention. In some instances, the absence of
mented globally during the period 1990-2018 were identified via interventions promoting private well maintenance behaviours
a systematised review and their results pooled with analyses delin- and risk awareness may be attributable to government capacity
eated to control for intervention characteristics and country devel- or lack of subsequent ‘‘follow through”. Bundschuh & Garcia
opment status (Schardt et al., 2007). Overall, findings indicated (2008) and McClintock et al. (2012) note a paucity of data regard-
that instructor-led, practical interventions allied with both large- ing current groundwater arsenic burdens and an absence of reme-
and local-scale awareness-raising campaigns may represent an dial interventions in Latin American countries; this in spite of
optimum approach for future private well risk prevention initia- recent estimates placing the number of Latin American populations
tives. Further to identifying the current success (and predictors exposed to chronic concentrations of arsenic at 4.5 million. Signif-
thereof) of pertinent risk mitigation strategies, this study high- icantly, Bundschuh & Garcia (2008) reference a small number of
10 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338

Table 8
Bivariate analysis between intervention outcome efficacy and intervention characteristics.

One-way One-way ANCOVA One-way ANCOVA One-way ANCOVA

ANOVA (Model 1) (Model 2) (Model 3)
F-statistic P value F-statistic P value F- P F- P
statistic value statistic value
Behaviour
Primary intervention 1.974 0.103 1.910 0.116 2.040 0.095 1.949 0.110
type
>1 intervention type 0.358 0.554 0.197 0.660 0.057 0.813 0.043 0.837
Intervention reach 1.066 0.356 0.986 0.385 1.328 0.280 1.236 0.305
Convenience 0.114 0.738 0.034 0.856 0.023 0.881 0.025 0.877
mechanism
Electronic mechanism 1.241 0.273 2.400 0.131 1.933 0.174 2.693 0.111
Interpersonal 1.218 0.278 1.299 0.263 0.907 0.348 1.028 0.318
mechanism
Print mechanism 0.040 0.843 0.020 0.887 0.039 0.844 0.140 0.710
Incentive 0.247 0.623 0.210 0.650 0.035 0.852 0.055 0.816
Edu/res. Coordinator 5.712 0.023 * 4.989 0.033 * 5.575 0.024 5.215 0.029
* *
Govt. body 0.566 0.457 0.536 0.469 0.304 0.585 0.340 0.564
coordinator
NPO coordinator 1.882 0.179 1.673 0.205 1.192 0.283 1.261 0.270
Knowledge
Primary intervention 1.743 0.179 1.734 0.184 1.662 0.201 1.670 0.202
type
>1 intervention type 0.205 0.655 0.084 0.775 0.377 0.546 0.277 0.605
Intervention reach 1.724 0.204 1.823 0.189 1.807 0.191 1.777 0.198
Convenience 0.764 0.392 0.502 0.487 1.846 0.189 1.599 0.221
mechanism
Electronic mechanism 0.136 0.716 0.560 0.463 0.133 0.719 0.486 0.494
Interpersonal 0.898 0.354 0.658 0.427 1.013 0.326 0.825 0.375
mechanism
Print mechanism 2.294 0.145 1.695 0.208 2.336 0.142 1.706 0.207
Incentive 0.000 0.996 0.033 0.858 0.001 0.981 0.006 0.938
Edu/res. Coordinator 0.495 0.489 0.323 0.576 0.497 0.489 0.343 0.565
Govt. body 0.464 0.503 0.518 0.480 0.621 0.440 0.918 0.350
coordinator
NPO coordinator 0.000 0.997 0.008 0.931 0.001 0.971 0.005 0.944

Differences in outcome efficacy by primary intervention type and intervention reach were analysed using One-way ANOVA.
Model 2: One-way ANCOVAs examined differences in outcome controlling for study design across listed variables.
Model 3: One-way ANCOVAs examined differences in outcome controlling for country development status across listed variables.
Model 4: One-way ANCOVAs examined differences in knowledge and outcome controlling for study design and country development status across listed variables.
* Significant at <0.05 level.

field studies piloting arsenic removal technologies. Although a pertaining to the rates and consequent risks of regional microbial
number of similar studies were identified in the current literature groundwater contamination become more widespread, a concur-
review (e.g. Cheng et al., 2004; Berg et al., 2006), such interven- rent increase in correspondent interventions is required to address
tions were evaluated based on contaminant removal efficiency increasing waste production and climate-related challenges to
and human health improvements as opposed to behaviour and groundwater quality. Although a number of studies external to this
knowledge attainment. As noted by Jones-Hughes et al. (2013), review have indeed evaluated interventions implemented in areas
interventions premised on installation of contaminant removal characterised by microbial groundwater contamination (Odagiri
technologies may not lead to sufficient, long-term health outcomes et al., 2016), outcomes were premised on improvements in human
in and of themselves and must be better tailored towards (or health and/or contaminant reduction as opposed to behaviour and
reported with respect to) behaviour and knowledge acquisition. knowledge attainment.

5.2. Reported contaminants 5.3. Intervention types

Notably, the majority of included studies evaluated interven- As is common in awareness raising interventions intended to
tions associated with reducing geogenic (chemical) contaminants reach relatively large audiences (Atkin & Rice, 2012), information
in specific ‘‘hotspots.” The most represented countries in the cur- campaigns were the most frequently reported intervention type
rent review, Bangladesh and the United States, are the subject of in both development contexts. Interventions in developing coun-
historically intensive groundwater arsenic mitigation programmes tries were typically associated with conventional water and sanita-
(Anderson et al., 2015; Henry & Suk, 2018). Conversely, and of tion initiatives often observed in low/medium-development
potential concern, prevention of microbial contamination was regions such as water testing services and provision of water treat-
encountered within just one identified intervention (Lule et al., ment mechanisms (Mosler, 2012). A greater number of workshop-
2005). This suggests a global lack of appreciation of the scale and based initiatives were noted in developed countries in the form of
severity of groundwater contamination by enteric pathogens and network-based forms of rural outreach delivered by state universi-
pervasive deficit of remedial interventions (Howard et al., 2006; ties in partnership with federal, state and local governments
Bradford & Harvey, 2017). While the global groundwater arsenic (Franz, 2014). However, interventions in developed countries were
burden is considerable (Ravenscroft et al., 2009), 35-59 million generally less predicated on direct engagement with and actions
annual cases of enteric infection are attributable to microbial on behalf of households, with water testing services and provision
groundwater contamination (Murphy et al., 2017). As information of treatment mechanisms less frequent. Additionally, interventions
 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338 11

Table 9 Table 9 (continued)

List of variable themes and corresponding types.
Variable Theme Variable Type
Variable Theme Variable Type
Random sample method (no/yes) Dichotomous*
Study location Representative sample (no/yes) Dichotomous*
Country name Categorical
Intervention outcome
Country development status Categorical
Behavioural outcome measure (no/yes) Dichotomous*
Continent Categorical
Contaminant outcome measure (no/yes) Dichotomous*
Study area scale Categorical
Health outcome measures (no/yes) Dichotomous*
Study area population Categorical
Knowledge outcome measure (no/yes) Dichotomous*
Study area geology Categorical
Behavioural outcome Dichotomous*
Contaminant(s) Contaminant outcome Dichotomous*
Total no. Numerical Health outcome Dichotomous*
Chemical (no/yes) Dichotomous* Knowledge outcome Dichotomous*
Microbial (no/yes) Dichotomous* Significant variable Categorical
Anthropogenic source (no/yes) Dichotomous*
*
Geogenic source (no/yes) Dichotomous* Dummy variable.

Human health
Disease(s)/illness(es) reported (no/yes) Dichotomous in developed countries adopted convenience mechanisms and
No. of disease(s)/illness(es) Numerical incentives significantly more often than developing countries.
Type of disease(s)/illness(es) Categorical
These findings indicate that private well interventions in devel-
Private groundwater oped countries may assume an appreciably greater capacity for pri-
Drilled well (no/yes) Dichotomous*
vate well stewardship and maintenance actions among target
Dug well (no/yes) Dichotomous*
Spring (no/yes) Dichotomous*
audiences. Conversely, opportunities for interpersonal engagement
Supply depth Categorical in developing nations may be more frequent due to less secure
Supply figures Categorical land title and stigma surrounding private well maintenance
Regulation (no/yes) Dichotomous (Boyle et al., 2010). It is thus advisable for intervention coordina-
Regulation type Categorical
tors in developed countries to engender trust and cooperation with
General intervention attributes communities from the outset and give greater consideration to the
Intervention type I Categorical
efficacy of convenience, incentive-based approaches (Re, 2015;
intervention type II Categorical
intervention type III Categorical Morris et al., 2016).
Intervention type IV Categorical
Reach Categorical
Duration in months Numerical
5.4. Intervention stakeholders
Cost stated (no/yes) Dichotomous*
Intervention activities
While the type of documented stakeholders within interven-
Convenience mechanism (no/yes) Dichotomous* tions was not strictly geographically orientated, participation of
No. of convenience mechanisms Numerical non-profit bodies was significantly related to development status.
Electronic mechanism (no/yes) Dichotomous* Moreover, downstream involvement of community actors (e.g.
No. of electronic mechanisms Numerical
assistance in information dissemination) was limited to eight
Interpersonal mechanism (no/yes) Dichotomous*
No. of interpersonal mechanisms Numerical interventions, all of which occurred in developing countries. The
Print mechanism (no/yes) Dichotomous* absence of documented community actors in developed countries
No. of print mechanisms Numerical may be seen as noteworthy, given that more than half of interven-
Total no. of mechanisms Numerical tions in developed countries were implemented at the regional- or
Incentive(s) (no/yes) Dichotomous
No. of incentives Numerical
local scale; interventions undertaken on a sub-national scale may
be better placed to utilise community channels to engage private
Intervention actors
Community body coordinator (no/yes) Dichotomous*
well uses on a more concerted, face-to-face basis (Morris et al.,
Educational/research body coordinator (no/yes) Dichotomous* 2016). Furthermore, community and civil society organisations
Government body coordinator (no/yes) Dichotomous* have been repeatedly recognised as an important means of private
Non-profit body coordinator (no/yes) Dichotomous* well user outreach in these regions (Morris et al., 2016; Henry &
Other type coordinator (no/yes) Dichotomous*
Suk, 2018). Significantly, the most frequent coordinator type docu-
No. of coordinators Numerical
Community body stakeholder (no/yes) Dichotomous* mented in developed countries was a government body, suggesting
Educational/research body stakeholder (no/yes) Dichotomous* that private groundwater interventions in developed countries are
Government body stakeholder (no/yes) Dichotomous* generally top-down in nature and may be insensitive to place-
Non-profit body stakeholder (no/yes) Dichotomous* specific circumstances. Educational/research bodies were the only
Other type stakeholder (no/yes) Dichotomous*
consistent coordinator type active in both development categories,
No. of stakeholders Numerical
suggesting a higher comparable level of expertise and potentially
Intervention evaluation
No. of evaluation methods Numerical
larger funding streams relative to other coordinators with respect
Clinical evaluation (no/yes) Dichotomous* to private well user outreach. As educational/research body coordi-
Survey evaluation (no/yes) Dichotomous* nators were associated with higher behavioural efficacy, they may
Water sample evaluation (no/yes) Dichotomous* be best placed to facilitate greater engagement and participation
Pre-intervention period measured (no/yes) Dichotomous*
with local, community-based organisations.
During intervention period measured (no/yes) Dichotomous*
Post-intervention period measured (no/yes) Dichotomous*
Pre-intervention period duration in months Numerical 5.5. Intervention outcomes
During intervention period duration in months Numerical
Post-intervention period duration in months Numerical
Convenience sample method (no/yes) Dichotomous* Although interventions undertaken in developed countries reg-
Purposive sample method (no/yes) Dichotomous* istered a slightly higher overall success rate than developing coun-
12 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338

Figure 4. Forest plot demonstrating levels of behavioural attainment (%) with corresponding intervention mechanisms.

Figure 5. Forest plot demonstrating levels of knowledge attainment (%) with corresponding intervention mechanisms.

tries (47% vs. 44%), mean behavioural outcome was analogous success rate in promoting behaviour change compared to informa-
across development type (47% vs. 48%). This may be partly ascribed tion campaigns and well testing services indicates the high produc-
to higher baseline rates of well maintenance (potentially related to tivity of first-hand, interactive interventions in stimulating
higher mean levels of education and income) and/or a lack of assis- appropriate private well maintenance. Notwithstanding the valid-
tance relative to interventions in developing countries (Flanagan ity of these interventions, it is vital that circumstantial and
et al., 2015). This low rate of behaviour change is somewhat similar audience-related factors are acknowledged in intervention design
to previous figures relating to private well user maintenance as standardised approaches may disregard social, value-based
actions measured in KAP surveys in developed countries, which and experiential factors such as frequent correlates with health
noted acceptable well maintenance behaviours in approximately and water maintenance behaviours (Figueroa & Kincaid, 2010;
30% of surveyed well users (Kreutzwiser et al., 2011; Malecki Moore & Boldero, 2017). As participation of educational/research
et al., 2017). Highest rates of behaviour change were associated coordinators was also shown to engender significantly higher rates
with safe source installation (mean improvement = 135%), of behaviour change (in both general terms and controlling for
although workshops (mean improvement = 65%) were the most development status and study design), their enhanced involve-
successful primary intervention type in this regard that did not ment in subsequent interventions may bear consideration at gov-
automatically stipulate direct assistance such as well installation ernment level. Involvement of intervention coordinators
or maintenance measures on behalf of householders. The greater combining pedagogical and technical expertise may increase the
 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338 13

likelihood of desired behavioural adaption on the part of well own- been undertaken in areas encompassing peri-urban municipalities,
ers, validating calls for greater involvement of educational experts in spite of specification of broad rurality. In light of these factors, it
in hydrogeological interventions (Re, 2015; Hynds et al., 2018). is urged that review findings, notwithstanding their potential
The lower increase in mean knowledge (47%) attained in devel- value, be utilised cautiously.
oping countries relative to behaviour may be attributable to lower
audience literacy rates and a less immediate requirement for infor-
mation assimilation by virtue of intervention type (e.g. safe source 6. Conclusion
installation undertaken by intervention coordinators/actors) (Jones
et al., 2013). Locally adopted interventions attained the highest Contamination of potable groundwater poses a significant glo-
improvement in behaviour while large-scale interventions attained bal health threat – one most acutely experienced among rural pri-
the highest knowledge increase with respect to audience reach. vate owners due to infrequent and inconsistent supply regulation,
Highest knowledge outcomes were attained by workshops, further localised natural and anthropogenic hazards and increased suscep-
underscoring the efficacy of workshops relative to other interven- tibility to contamination pathways shaped by modern peri-urban/
tion types. Two cooperative extension programmes predicated on rural development and climate change dynamics. The pervasive
interpersonal workshops and convenience-based measures (e.g. absence of ‘‘top-down” supply management via regulation and
provision of water sample bottle, well test result interpretation) consequent burden of responsibility placed on well owners them-
registered notably high behaviour outcomes (Benham et al., selves prescribes an evidence-based, communicative approach to
2016; Clemens et al., 2007). However, as was also observed for groundwater risk prevention. The lack of reported interventions
behaviour, type of intervention mechanism demonstrated no sig- in Europe, Australia and South America underscores the potential
nificant relationship with knowledge outcome. This may denote significance of ‘‘place-based” factors such as social geography, pol-
that efficacy of intervention activities is premised on the quality icy and institutional capacity/research priorities in determining the
as opposed to type of engagement material, as noted by Figueroa occurrence and evaluation of private well interventions. While dif-
& Kincaid (2010) in the context of developing nations. Notably, ferent geographical regions may require different approaches, the
the majority of knowledge measures were based on awareness of absence of community and voluntary sector coordinators in inter-
contamination risk as opposed to knowledge of private well main- ventions within developed countries set against requests for their
tenance; attainment of 1 form of supply maintenance knowledge increased involvement therein points to a disconnect that requires
was measured in only two interventions, once more evincing a lack attention in intervention design/strategies. This disconnect is also
of focus on well user self-efficacy (Hynds et al., 2018). manifest in the distinct lack of private well interventions pertain-
The viability of locally scaled interventions in addressing com- ing to microbial contamination, which warrants increased consid-
plex human health scenarios (e.g. contaminated domestic water eration by governmental and hydrogeological communities alike.
supplies) is repeatedly cited in the health outreach literature and Differences in intervention type (and performance/outcome)
further reinforces the efficacy of locally tailored interventions in with regard to development status may be anticipated given the
promoting enhanced private groundwater quality, stewardship socio-economic range of target audiences and consequently vary-
and user health (Craig et al., 2008; Figueroa & Kincaid, 2010). Con- ing intervention priorities. However, while there is inherent value
versely, supra-regional (i.e. state/national) interventions such as in examining development categories separately, comparisons may
public communication campaigns may also be required to raise ini- be drawn. Interventions undertaken in developing countries are
tial awareness or generate discourse; large-scale campaigns per- distinguished by a stronger local focus and a more interpersonal,
formed markedly better than medium-scale interventions. The incentivised-approach. Given the higher overall behaviour rate
efficacy of source remediation and safe source installation cam- exhibited among developing country interventions and the
paigns and workshops suggests that a combination of household observed efficacy of locally-implemented initiatives, an informed
or community-based interventions with interpersonal, practical adoption of comparable measures in developed countries may
interventions represent optimal intervention strategies. Such result in comparable outcomes. However, higher knowledge rate
interventions are inherently more effective at the local level but increases associated with large-scale interventions may indicate
may transcend individual town or village setting if replicated con- the value of a wide geographical reach in generating greater collec-
sistently in accordance with an overarching strategy and audience tive discussion and by extension awareness. While the docu-
analysis (Milat et al., 2015). On this basis, it is recommended that mented efficacy of education/research coordinator-led
groundwater outreach practitioners prioritise design and employ- interventions and workshops reinforces the validity of instructor-
ment of both locally focused, interdisciplinary rural private well led interventions, it is nevertheless vital that they are tailored
interventions and large-scale awareness raising strategies, allied accordingly based on circumstantial factors and audience analysis.
with educational initiatives and tailored awareness-raising cam-
paigns. Concerted and sustained collaboration among stakeholders Conflict of Interest
comprising hydrogeologists, behavioural scientists and commu-
nity/ voluntary agencies will be necessary to encourage develop- All authors confirm that no actual or potential conflict of inter-
ment of feasible, replicable intervention frameworks and, by est exists including any financial, personal or other relationships
extension, prevalent rural private groundwater stewardship. with other people or organizations within three years of beginning
the submitted work that could inappropriately influence, or be per-
5.6. Study limitations ceived to influence, this work.

The current review strictly adhered to a systematised review

framework and is thus characterised by a number of inherent lim- References
itations. As with all scientific reviews, the potential presence of
Anderson, B.E., Naujokas, M.F., Suk, W.A., 2015. Interweaving knowledge resources
publication bias may impact on the type and location of studies to address complex environmental health challenges. Environmental Health
published and reported outcomes. Interventions were largely con- Perspectives 123, 1095–1099.
centrated within two continental sub-regions, thus potentially lim- Andrade, L., O’Dwyer, J., E., O’Neill, E. & Hynds, P., 2018. Surface water flooding,
groundwater contamination, and enteric disease in developed countries: a
iting the transferability of intervention findings on a global scale. It scoping review of connections and consequences. Environmental pollution 236,
is further anticipated that a number of interventions may have 540–549.
14 S. Mooney et al. / Science of the Total Environment 716 (2020) 135338

Atkin, C.K., Rice, R.E., 2012. Theory and Principles of Public Communication constraining private water well stewardship. Journal of Environmental
Campaigns. In: Rice, R.E., Atkin, C.K. (Eds.), Public Communication Campaigns. Management 92, 1104–1114.
4th ed. Thousand Oaks, CA, SAGE. Lapworth, D.J., Baran, N., Stuart, M.E., Ward, R.S., 2012. Emerging organic
Benham, B., Ling, E., Ziegler, P., Krometis, L.A., 2016. Whatâ€TMs in Your Water? contaminants in groundwater: A review of sources, fate and occurrence.
Development and Evaluation of the Virginia Household Water Quality Program Environmental Pollution 163, 287–303.
and Virginia Master Well Owner Network. Journal of Human Sciences and Lule, J.R., Mermin, J., Ekwaru, J.P., Malamba, S., Downing, R., Ransom, R., et al., 2005.
Extension 4 (1), 123–138. Effect of home-based water chlorination and safe storage on diarrhea among
Bradford, S.A., Harvey, R.W., 2017. Future research needs involving pathogens in persons with human immunodeficiency virus in Uganda. American Journal of
groundwater. Hydrogeology Journal 25 (4), 931–938. Tropical Medicine and Hygiene 73, 926–933.
Berg, M., Luzi, S., Trang, P.T.K., Viet, P.H., Giger, W., Stüben, D., 2006. Arsenic Malecki, K.M.C., Schultz, A.A., Severtson, D.J., Anderson, H.A., VanDerslice, J.A., 2017.
Removal from Groundwater by Household Sand Filters: Comparative Field Private-well stewardship among a general population based sample of private
Study, Model, Calculations, and Health Benefits. Environmental Science & well-owners. Science of the Total Environment 601–602, 1533–1543.
Technology 40, 5567–5573. Milat, A.J., Bauman, A., Redman, S., 2015. Narrative review of models and success
Boyle, K.J., Kuminoff, N.V., Zhang, C., Devanney, M., Bell, K.P., 2010. Does a property- factors for scaling up public health interventions. Implementation Science 10, 113.
specific environmental health risk create a ‘‘neighborhood” housing price Mitchell, M., Curtis, A., Sharp, E., Mendham, E., 2012. Directions for social research
stigma? Arsenic in private well water. Water Resources Research 46, W03507. to underpin improved groundwater management. Journal of Hydrology 448–
Bunn, M.D., Savage, G.T., Holloway, B.B., 2002. Stakeholder analysis for multi-sector 449, 223–231.
innovations. The Journal of Business & Industrial Marketing 17 (2/3), 181–203. Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., 2009. Preferred Reporting Items for
Charrois, J.W.A., 2010. Private Drinking Water Supplies: Challenges for Public Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Medicine
Health. Canadian Medical Association Journal 182 (10), 1061–1064. 6, (7) e100009.
Cheng, Z., van Geen, A., Jing, C., Meng, X., Seddique, A., Ahmed, K.M., 2004. Moore, H.E. & Boldero, J. (2017) Designing Interventions that Last: A Classification of
Performance OF A Household-Level Arsenic Removal System during 4-Month Environmental Behaviors in Relation to the Activities, Costs, and Effort Involved
Deployments in Bangladesh. Environmental Science & Technology 38, 3442– for Adoption and Maintenance. Frontiers in Psychology, 8: 1874.
3448. Morris, L., Wilson, S., Kelly, W., 2016. Methods of conducting effective outreach to
Clemens, S.S., Swistock, B.R., Sharpe, W.E., 2007. The master well owner network: private well owners – a literature review and model approach. Journal of Water
volunteers educating Pennsylvania well owners. J. Extension 45 (4). and Health 14 (2), 167–182.
Craig, P., Dieppe, P., Macintyre, S., Michie, S., Nazareth, I., Petticrew, M., 2008. Mosler, H.J., 2012. A systematic approach to behavior change interventions for the
Developing and evaluating complex interventions: the new Medical Research water and sanitation sector in developing countries: a conceptual model, a
Council guidance. BMJ 337, a1655. review, and a guideline. International Journal of Environmental Health Research
Elamin, M.B., Flynn, D.N., Bassler, D., Briel, M., Alonso-Coello, P., Karanicolas, P.J., 22 (5), 431–449.
Guyatt, G.H., Malaga, G., Furukawa, T.A., Kunz, R., Schünemann, H., Murad, M.H., Murphy, H.M., Prioleau, M.D., Borchardt, M.A., Hynds, P.D., 2017. Review:
Barbui, C., Ciprianij, A., Montori, V.M., 2009. Choice of data extraction tools for Epidemiological evidence of groundwater contribution to global enteric
systematic reviews depends on resources and review complexity. Journal of disease, 1948–2015. Hydrogeology Journal 25 (4), 981–1001.
Clinical Epidemiology 62, 506–510. Niu, B., Loáiciga, H.A., Wang, Z., Benjamin Zhan, F., Hong, S., 2014. Twenty years of
Falagas, M.E., Pitsouni, E.I., Malietzis, G.A., Pappas, G., 2008. Comparison of PubMed, global groundwater research: A Science Citation Index Expanded-based
Scopus, Web of Science, and Google Scholar: strengths and weaknesses. The bibliometric survey (1993–2012). Journal of Hydrology 519 (A), 966–975.
FASEB Journal 22 (2), 338–342. Odagiri, M., Schriewer, A., Daniels, M.E., Wuertz, S., Smith, W.A., Clasen, T., . . .
Figueroa, M.E., Kincaid, D.L., 2010. Social, Cultural and Behavioral Correlates of Panigrahi, P., 2016. Human fecal and pathogen exposure pathways in rural
Household Water Treatment and Storage. Center Publication HCI 2010–1: Health Indian villages and the effect of increased latrine coverage. Water research 100,
Communication Insights. Johns Hopkins Bloomberg School of Public Health. 232–244.
Center for Communication Programs, Baltimore, MD. O’Dwyer, J., Morris Downes, M., Adley, C.C., 2016. The impact of meteorology on the
Franz, N.K., 2014. Measuring and Articulating the Value of Community Engagement: occurrence of waterborne outbreaks of vero cytotoxin-producing Escherichia
Lessons Learned from 100 Years of Cooperative Extension Work. Journal of coli (VTEC): a logistic regression approach. Journal of Water and Health 14 (1),
Higher Education Outreach and Engagement 18 (2), 5–16. 39–46.
Flanagan, S.V., Marvinney, R.G., Johnston, R.A., Yang, Q., Zheng, Y., 2015. Oxman, A.D., Lavis, J.N., Lewin, S. & Fretheim, A. (2009) SUPPORT Tools for evidence-
Dissemination of well water arsenic results to homeowners in Central Maine: informed health Policymaking (STP) 10: Taking equity into consideration when
influences on mitigation behavior and continued risks for exposure. Science of assessing the findings of a systematic review. Health Research Policy and
the Total Environment 505, 1282–1290. Systems, 7(1): S10.
Gleeson, T., Befus, K.M., Jasechko, S., Luijendijk, E., Bayani Cardenas, M., 2016. The Paul, M.P., Rigrod, P., Wingate, S., Borsuk, M.E., 2015. A community-driven
global volume and distribution of modern groundwater. Nature Geoscience 9, intervention in Tuftonboro, New Hampshire, succeeds in altering water
161–167. testing behavior. Journal of environmental health 78 (5), 30.
Gorelick, S.M., Zheng, C., 2015. Global change and the groundwater management Pearson, M., Jones-Hughes, T., Whear, R., Cooper, C., Peters, J., Evans, E.H., Depledge,
challenge. Water Resources Research 51, 3031–3051. M., 2011. Are interventions to reduce the impact of arsenic contamination of
Green, T.R., Taniguchi, M., Kooi, H., Gurdak, J.J., Allen, D.M., 2011. Beneath the groundwater on human health in developing countries effective?: a systematic
surface of global change: Impacts of climate change on groundwater. Journal of review protocol. Environmental Evidence 1, 1.
Hydrology 405, 532–560. Ravenscroft, P., Brammer, H., Richards, K., 2009. Arsenic Pollution: A Global
Haddaway, N.R., Collins, A.M., Coughlin, D., Kirk, S., 2015. The Role of Google Scholar Synthesis. John-Wiley & Sons, Oxford.
in Evidence Reviews and Its Applicability to Grey Literature Searching. PLoS Re, V., 2015. Incorporating the social dimension into hydrogeochemical
ONE 10, (9) e0138237. investigations for rural development: the Bir Al-Nas approach for socio-
Henry, H.F., Suk, W.A., 2018. Public Health and Karst Groundwater Contamination: hydrogeology. Hydrogeology Journal 23 (7), 1293–1304.
From Multidisciplinary Research to Exposure Prevention. In: White, W.B., Rowe, G., Frewer, L.J., 2005. A Typology of Public Engagement Mechanisms. Science,
Herman, J.S., Herman, E.K., Rutigliano, M. (Eds.), Karst Groundwater Technology, & Human Values 30 (2), 251–290.
Contamination and Public Health. Basel, Springer Nature, pp. 7–14. Schardt, C., Adam, S., M.B., Owens, T., Keitz, S. & Fontelo, P. (2007). Utilization of the
Higgins, J.P.T., Green, S. (Eds.), 2011. Cochrane Handbook for Systematic Reviews of PICO framework to improve searching PubMed for clinical questions. BMC
Interventions, Version 5.1.0. Cochrane, London. Medical Informatics and Decision Making, 7: 16.
Howard, G., Bartram, J., Pedley, S., Schmoll, O., Chorus, I., Berger, P., 2006. Schwarzenbach, R.P., Egli, T., Hofstetter, T.B., von Gunten, U., Wehrli, B., 2010. Global
Groundwater and public health. In: Schmoll, O., Howard, G., Chilton, J., Water Pollution and Human Health. Annual Review of Environment and
Chorus, I. (Eds.), Protecting Groundwater for Health: Managing the Quality of Resources 35, 109–136.
Drinking-water Sources. IWA, London. Shankar, S., Shanker, U., Shikha, 2014. Arsenic Contamination of Groundwater: A
Hynds, P.D., Thomas, M.K., Pintar, K.D., 2014. Contamination of groundwater review of sources, prevalence, health risks, and strategies for mitigation. The
systems in the US and Canada by enteric pathogens, 1990‑‘2013: a review and Scientific World Journal 2014, 304524.
pooled-analysis. PLoS One. 9, (5) e93301. UNDP, 2016. Human Development Report 2016: Human Development for Everyone.
Hynds, P., Regan, S., Andrade, L., Mooney, S., O’Malley, K., DiPelino, S., O’Dwyer, J., United Nations Development Programme, New York City, NY.
2018. Muddy Waters: Refining the Way forward for the ‘‘Sustainability Science” Valle Junior, R.F., Varandas, S.G.P., Sanches Fernandes, L.F., Pachecho, F.A.L., 2014.
of Socio-Hydrogeology. Water 10 (9), 1111. Groundwater quality in rural watersheds with environmental land use conflicts.
Jakeman, A.J., Barreteau, O., Hunt, R.J., Rinaudo, J.D., Ross, A., Arshad, M., Hamilton, Science of the Total Environment 493, 812–827.
S., 2016. Integrated Groundwater Management: An Overview of Concepts and Waddington, H., White, H., Snilstveit, B., Garcia Hombrados, J., Vojtkova, M., Davies,
Challenges. In: Jakeman, A.J., Barreteau, O., Hunt, R.J., Rinaudo, J.D., Ross, A. P., Bhavsara, A., Eyers, J., Perez Koehlmoos, T., Petticrew, M., Valentine, J.C.,
(Eds.), Integrated Groundwater Management: Concepts, Approaches and Tugwell, P., 2012. How to do a good systematic review of effects in international
Challenges. Springer, New York City, NY. development: a tool kit. Journal of Development Effectiveness 4 (3), 359–387.
Jones, S.A., Sanford Bernhardt, K.L., Kennedy, M., Lantz, K., Holden, T., 2013. WHO, 2009. Practical Guidance for Scaling up Health Service Innovations. World
Collecting Critical Data to Assess the Sustainability of Rural Infrastructure in Health Organization, Geneva.
Low-Income Countries. Sustainability 5 (11), 4870–4888. WHO/UNICEF (2015) 25 Years Progress on Sanitation and Drinking Water: 2015
Kreutzwiser, R., de Loë, R., Imgrund, K., Conboy, M.J., Simpson, H., Plummer, R., Update and MDG Assessment. New York, NY: United Nations Children’s Fund
2011. Understanding stewardship behaviour: Factors facilitating and and World Health Organisation.