bmj open | a leading multidisciplinary medical journal - bmj ......distributed as the published...

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available. When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to. The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript. BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com). If you have any questions on BMJ Open’s open peer review process please email

[email protected]

on June 8, 2021 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2019-035143 on 5 May 2020. D

ownloaded from

http://bmjopen.bmj.com/[email protected]://bmjopen.bmj.com/

For peer review onlyImpact of a farmers' market nutrition coupon program on

diet quality and psychosocial well-being among low-income adults: Protocol for a randomized controlled trial and a

longitudinal qualitative investigation

Journal: BMJ Open

Manuscript ID bmjopen-2019-035143

Article Type: Protocol

Date Submitted by the Author: 20-Oct-2019

Complete List of Authors: Aktary, Michelle; University of Calgary Faculty of KinesiologyCaron-Roy, Stephanie ; University of Calgary Faculty of KinesiologySajobi, Tolu; University of Calgary Cumming School of Medicine, Department of Community Health SciencesO'Hara, Heather ; British Columbia Association of Farmers' MarketsLeblanc, Peter ; British Columbia Association of Farmers' MarketsDunn, Sharlette; University of Calgary Cumming School of Medicine, Department of Community Health SciencesMcCormack, Gavin R.; University of Calgary, Department of Community Health Sciences; University of Calgary Faculty of KinesiologyTimmins, Dianne; University of Calgary Cumming School of Medicine, Department of Community Health SciencesBall, Kylie; Deakin University, Centre for Physical Activity and Nutrition ResearchDowns, Shauna; Rutgers The State University of New Jersey, Minaker, Leia; University of Waterloo, Propel Centre for Population Health ImpactNykiforuk, Candace; University of Alberta School of Public HealthGodley, Jenny; University of Calgary, Department of SociologyMilaney, Katrina; University of Calgary Cumming School of Medicine, Department of Community Health SciencesLashewicz, Bonnie ; University of Calgary Cumming School of Medicine, Department of Community Health SciencesFournier, Bonnie ; Thompson Rivers University School of NursingElliott, Charlene ; University of Calgary Faculty of Kinesiology; University of Calgary Department of Communication Media and FilmRaine, Kim; University of Alberta School of Public HealthProwse, Rachel; University of Alberta School of Public HealthOlstad, Dana; University of Calgary Cumming School of Medicine, Department of Community Health Sciences; University of Calgary Faculty of Kinesiology

Keywords:PUBLIC HEALTH, QUALITATIVE RESEARCH, STATISTICS & RESEARCH METHODS, Health policy < HEALTH SERVICES ADMINISTRATION & MANAGEMENT, NUTRITION & DIETETICS

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

BMJ Open on June 8, 2021 by guest. P

rotected by copyright.http://bm

jopen.bmj.com

/B

MJ O

pen: first published as 10.1136/bmjopen-2019-035143 on 5 M

ay 2020. Dow

nloaded from

http://bmjopen.bmj.com/

For peer review only

Page 1 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from


For peer review onlyI, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance with the terms applicable for US Federal Government officers or employees acting as part of their official duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd (“BMJ”) its licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence.

The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to the Submitting Author unless you are acting as an employee on behalf of your employer or a postgraduate student of an affiliated institution which is paying any applicable article publishing charge (“APC”) for Open Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative Commons licence – details of these licences and which Creative Commons licence will apply to this Work are set out in our licence referred to above.

Other than as permitted in any relevant BMJ Author’s Self Archiving Policies, I confirm this Work has not been accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate material already published. I confirm all authors consent to publication of this Work and authorise the granting of this licence.

Page 2 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from

https://authors.bmj.com/wp-content/uploads/2018/11/BMJ_Journals_Combined_Author_Licence_2018.pdfhttp://creativecommons.org/http://bmjopen.bmj.com/


1

1 Impact of a farmers' market nutrition coupon program on diet quality and psychosocial well-being 2 among low-income adults: Protocol for a randomized controlled trial and a longitudinal qualitative 3 investigation45 Michelle L. Aktary1, Stephanie Caron-Roy1, Tolulope Sajobi2, Heather O'Hara3, Peter Leblanc3, Sharlette 6 Dunn2, Gavin R. McCormack1,2,4, Dianne Timmins2, Kylie Ball5, Shauna Downs6, Leia M. Minaker7, 7 Candace I. J. Nykiforuk8, Jenny Godley9, Katrina Milaney2, Bonnie Lashewicz2, Bonnie Fournier10, 8 Charlene Elliott1, 11, Kim D. Raine8, Rachel J. L. Prowse8, Dana Lee Olstad1, 2*9

101112 1Faculty of Kinesiology, University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada; 13 2Department of Community Health Sciences, Cumming School of Medicine, University of Calgary, 3280 14 Hospital Drive NW, Calgary, AB T2N 4N1, Canada; 3British Columbia Association of Farmers' Markets, 15 203-2642 Main Street, Vancouver, BC V5T 3E6, Canada; 4School of Architecture, Planning and 16 Landscape, University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada; 5Institute for 17 Physical Activity and Nutrition (IPAN), Deakin University, 75 Pigdons Rd Waurn Ponds, Geelong 3216, 18 Australia; 6School of Public Health, Rutgers University, 683 Hoes Lane W, Piscataway, NJ 08854, USA; 19 7School of Planning, University of Waterloo, 200 University Avenue West, Waterloo, ON N2L 3G1, 20 Canada; 8School of Public Health, University of Alberta, 3-300 Edmonton Clinic Health Academy, 11405-21 87 Ave Edmonton, AB T6G 1C9, Canada; 9Department of Sociology, University of Calgary, 2500 22 University Drive NW, Calgary, AB T2N 1N4, Canada; 10School of Nursing, Thompson Rivers University, 23 805 TRU Way, Kamloops, BC V2C 0C8, Canada; 11Department of Communication, Media and Film, 24 University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada252627 *Corresponding author:28 Dr. Dana Lee Olstad, Department of Community Health Sciences, Teaching, Research and Wellness 29 Building, 3280 Hospital Drive NW, Calgary, Alberta, Canada, T2N 4Z6 30 Email: [email protected], Phone: 403-210-8673313233 Word count: 7,941343536373839404142434445464748

Page 3 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from

mailto:[email protected]://bmjopen.bmj.com/


2

49 ABSTRACT

50 Introduction: Low-income populations have poorer diet quality and lower psychosocial well-being than

51 their higher income counterparts. These inequities increase the burden of chronic disease in low-income

52 populations. Farmers’ market subsidy programs may improve diet quality and psychosocial well-being

53 among low-income populations. In Canada, the British Columbia (BC) Farmers’ Market Nutrition Coupon

54 Program (FMNCP) aims to improve dietary patterns and health among low-income participants by

55 providing coupons to purchase healthy foods from farmers’ markets. This study will assess the impact of

56 the BC FMNCP on the diet quality and psychosocial well-being of low-income adults and explore

57 mechanisms of program impacts.

58 Methods and analysis: In a parallel group randomized controlled trial, low-income adults will be

59 randomized to a FMNCP intervention (n= 132) or a no-intervention control group (n= 132). The FMNCP

60 group will receive 16 coupon sheets valued at $21/sheet over 10-15 weeks to purchase fruits,

61 vegetables, dairy, meat/poultry/fish, eggs, nuts, and herbs at farmers’ markets and will be invited to

62 participate in nutrition skill-building activities. Overall diet quality (primary outcome), diet quality

63 subscores, mental well-being, sense of community, food insecurity, and risk of malnutrition (secondary

64 outcomes) will be assessed at baseline, immediately post-intervention, and 16-weeks post-intervention.

65 Dietary intake will be assessed using the Automated Self-Administered 24-hour Dietary Recall. Diet

66 quality will be calculated using the Healthy Eating Index-2015. Repeated measures mixed-effect

67 regression will assess differences in outcomes between groups from baseline to 16-weeks post-

68 intervention. Furthermore, 20-25 participants will partake in semi-structured interviews during- and 8-

69 weeks after program completion to explore participants’ experience with and perceived outcomes from

70 the program.

71 Ethics and dissemination: This study has received ethics approval from the Conjoint Health Research

72 Ethics Board of the University of Calgary (REB18-0508) and two other Boards. Findings will be

Page 4 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



3

73 disseminated through deliberative dialogue, policy briefs, lay summaries, conference presentations and

74 peer-reviewed publications.

75

76 Trial registration: ClinicalTrials.gov (NCT03952338), Registered May 10, 2019,

77 https://clinicaltrials.gov/ct2/show/NCT03952338?term=farmers%27+market&rank=2

7879 Key words: Farmers’ market, diet quality, low-income, 24-hour dietary recall, psychosocial, subsidy, 80 food insecurity, randomized controlled trial, longitudinal, qualitative81

82 Strengths and limitations of this study

83 1. This randomized controlled trial will provide the first assessment of the causal impact of a

84 farmers’ market healthy food subsidy on the diet quality and psychosocial well-being of low-

85 income adults and will provide evidence of the sustainability of program impacts.

86 2. This study will use valid measurement tools to assess outcomes, thus increasing accuracy of

87 effect estimates.

88 3. A longitudinal qualitative evaluation will explore participants’ experiences of accessing

89 nutritious foods, perceived outcomes and how they were achieved, to inform program

90 improvements.

91 4. The data are self-reported and, therefore, subject to self-reported measurement bias, and as

92 the study is longitudinal, there is also a risk of loss to follow-up.

93

94

95

96

97

98

Page 5 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



4

99 BACKGROUND

100 Income is amongst the strongest determinants of diet quality [1, 2] and overall health.[1] A clear

101 socioeconomic gradient exists whereby individuals with lower incomes experience higher rates of

102 nutrition-related chronic diseases [3-5] relative to those with higher incomes. Low household income is

103 also a key determinant of household food insecurity,[6-11] which is associated with lower diet quality

104 [12-15] and inadequate nutrient intake.[16, 17] Evidence suggests that low-income populations tend to

105 consume diets lower in fruits and vegetables and higher in refined white grains, high-fat meats, fried

106 foods and added fats.[18, 19] These inequities in diet quality may partly explain the greater vulnerability

107 of low-income and food insecure populations to poor health outcomes [18, 20-22] and

108 undernutrition.[16, 17] Therefore, it is crucial that public health interventions seek to support healthy

109 dietary patterns among low-income populations.

110

111 Determinants of diet quality are complex and multifactorial.[23] One approach to conceptualizing the

112 many factors underpinning differential dietary patterns among low-income groups is through the

113 socioecological model.[24, 25] The socioecological model depicts the complex and reciprocal

114 interactions among multiple levels of influence, including individual, social, community, and policy level

115 factors that shape dietary patterns and health outcomes.[23-26] At the individual level, factors such as

116 psychological state [19, 27] and nutrition-related knowledge [27] have been shown to influence dietary

117 patterns.[26, 28-31] For instance, studies have demonstrated that high self-efficacy for consuming fruits

118 and vegetables [32, 33] is associated with greater fruit and vegetable intake. The social level

119 encompasses social and cultural contexts that influence dietary patterns.[27, 34] Low-income

120 populations generally have lower social support [33, 35] and social capital [36-38] compared to higher

121 income populations,[36, 39, 40] which are, in turn, associated with poorer dietary intake and health

122 outcomes.[33, 41] The community level includes the physical environments in which people live and

Page 6 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



5

123 work.[25, 31] Studies from the United States (U.S.)[25, 42-45] and Canada [46] have shown that

124 disadvantaged neighbourhoods generally have more fast food outlets, which are associated with greater

125 purchasing and consumption of unhealthy foods.[43] The policy level encompasses policies and

126 regulations that influence the social distribution of dietary patterns and health outcomes across a

127 population.[1, 47] Interventions at the policy level provide significant potential for sustainable, cost-

128 effective and equitable health impacts.[48, 49] Fiscal policies that influence food prices and affordability

129 (e.g., taxation, subsidies) are particularly important for supporting healthy dietary patterns among low-

130 income groups,[50] because the economic resources of low-income populations are often insufficient to

131 purchase healthy foods consistent with dietary recommendations.[18, 51, 52]

132

133 Given the multi-level contexts in which dietary patterns are situated, interventions at any single level

134 (e.g., individual level interventions such as nutrition education) are unlikely to substantially improve diet

135 quality and health outcomes among low-income populations.[25, 53] Multi-sectoral collaboration is

136 required to develop and implement policies and programs that address determinants of poor dietary

137 patterns and ill health at all levels [54-56] to effectively reduce inequities in dietary patterns and

138 improve health outcomes.[57, 58]

139

140 Farmers’ markets, described as retail spaces where farmers and other growers gather to sell fresh,

141 locally grown foods directly to customers,[59] are growing in interest as multi-component interventions

142 that aim to improve access to and intake of nutritious foods and reduce food insecurity among low-

143 income populations.[60-63] Several U.S. studies have demonstrated an association between provision of

144 farmers’ market fruit and vegetable subsidies and increased fruit and vegetable intake in low-income

145 populations.[60, 64-68] These policy level, government funded subsidy programs primarily aim to

146 reduce the financial burden associated with procuring healthy, locally-sourced produce.[61] However,

Page 7 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



6

147 they also influence factors situated at the individual, community, and social level of the socioecological

148 model. At the community level, farmers’ markets have potential to alleviate barriers associated with

149 accessing healthy foods,[45, 69, 70] as they offer fresh, local produce [45, 59, 71] and can be set up in

150 communities that otherwise have limited access to healthy foods.[67] Farmers’ market food subsidies

151 may, therefore, encourage the development of farmers’ markets in low-income neighbourhoods.[72]

152 Farmers’ market food subsidies also support local farmers and promote sustainable local food systems

153 [73] by increasing awareness of farmers’ markets within communities [74] and increasing the customer

154 base, thereby generating increased farmers’ market sales.[75] Farmers’ market food subsidy programs

155 can also influence social and individual level determinants of dietary intake.[32, 76] Farmers’ markets

156 act as social spaces, facilitating knowledge exchange between customers and farmers, and increasing

157 social interactions between community members.[59, 77] These social aspects of farmers’ markets are

158 particularly important for the well-being of low-income groups, as social exclusion and isolation are

159 common side effects of food insecurity and low-income status.[78] Finally, at the individual level,

160 farmers’ market programs that offer nutrition skill-building activities may enhance participant food- and

161 nutrition-related knowledge and skills [32, 79] and attitudes towards the importance of fruit and

162 vegetable consumption.[79]

163

164 Farmers’ market food subsidy programs may, therefore, represent a promising multi-level approach to

165 improve the diet quality and psychosocial well-being [80] of low-income populations; however, several

166 knowledge gaps remain [67]. Notably, most previous studies have been cross-sectional [65, 66, 69, 81-

167 84] or used a pre/post design,[60, 63, 85] and/or lacked a control group,[63, 69, 86] each of which does

168 not allow for causal inference.[67] Randomized controlled trials (RCT) or longitudinal studies can provide

169 stronger evidence of the dietary and health impacts of farmers’ market food subsidy programs.[67] In

170 addition, most studies have only sought to assess change in fruit and vegetable consumption, and have

Page 8 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



7

171 primarily measured dietary intake by using brief fruit and vegetable screeners [67] rather than more

172 comprehensive and valid assessment tools such as food frequency questionnaires and 24-hour dietary

173 recalls.[67, 74] Assessment of overall dietary intake is important, as when one aspect of diet changes,

174 such as fruit and vegetable intake, concurrent changes occur in other aspects of dietary intake that must

175 be considered.[87] Furthermore, the majority of studies have been conducted in the U.S., and evidence

176 from other nations is sparse.[45]

177

178 Efforts to more fully understand the impacts of farmers’ market food subsidy programs will be enhanced

179 by investigating participants’ experiences of accessing nutritious foods and their perceptions of

180 outcomes from these programs. Qualitative studies have shown positive perceived outcomes from

181 farmers’ market food subsidy programs, including perceived greater exposure to, and intake of, fruits

182 and vegetables; increased resources to purchase healthy foods;[88, 89] and improved quality of life and

183 mental well-being.[90] However, studies examining participants’ experiences have also identified

184 barriers to visiting farmers’ markets, such as lack of access to transportation, limited market hours, and

185 the perception that farmers’ markets are more expensive than other retailers.[88, 89] In addition, the

186 use of subsidies in farmers’ markets may set participants apart from the general farmers’ market

187 customer population, leading to feelings of stigma.[89] Most previous qualitative studies are limited by

188 investigations conducted at a single point in time,[76, 88-90] which also limits understanding of how

189 participants’ experiences change once subsidy programs end and whether outcomes are maintained

190 over time. Understanding participants’ experiences and perceived outcomes from farmers’ market food

191 subsidy programs over time can help identify mechanisms of program impact and inform policies and

192 interventions that seek to address health inequities.[91]

193

Page 9 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



8

194 In British Columbia (BC), Canada, the average monthly cost to purchase a healthy diet for a family of four

195 is $1,019,[92] nearly one-half the income from low-wage employment.[93] The BC Farmers’ Market

196 Nutrition Coupon Program (FMNCP) is a healthy eating initiative that offers a healthy food subsidy,

197 along with supportive nutrition skill-building activities, for low-income populations.[94] The BC FMNCP

198 began in 2007 in response to a longstanding interest in connecting farmers’ markets with programs that

199 support the nutritional needs of low-income households.[94] It is the only government funded program

200 of this type in Canada. The program facilitates access to nutritious foods for low-income families,

201 pregnant women, and older adults by providing participants with coupons valued at $21/week to

202 purchase fruits, vegetables, dairy, meat/poultry/fish, eggs, nuts, and cut herbs from participating BC

203 farmers’ markets.[94] The goal of the BC FMNCP is to provide financial support for low-income

204 households to purchase and consume healthier foods, thereby improving diet quality [53, 95] and

205 overall health.[96, 97] The program also aims to minimize further marginalization of low-income

206 individuals by encouraging their participation in farmers’ markets, which are important social spaces

207 that may foster social and mental well-being.[61, 98, 99] Currently, the FMNCP operates in 57

208 communities across BC and reaches over 3900 households;[94] however, the need remains substantial,

209 with > 15 communities on waiting lists to participate in the program. It is unclear if the BC FMNCP is

210 achieving its aims, as the program’s outcomes have not been rigorously investigated.

211

212 This study was co-designed with stakeholders from the BC Association of Farmers’ Market and the

213 FMNCP in order to achieve the following objectives:

214 1) Conduct a RCT to investigate the impact of the BC FMNCP on:

215 a. overall diet quality (primary outcome),

216 b. diet quality subscores, mental well-being, sense of community, experiences of food

217 insecurity, risk of malnutrition (secondary outcomes), and

Page 10 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



9

218 c. subjective social status (exploratory outcome) of low-income adults.

219 2) Conduct a longitudinal qualitative evaluation to:

220 a. describe participants’ experiences of accessing nutritious foods, including facilitators

221 and barriers, during and after the BC FMNCP, and

222 b. explore perceived short-term outcomes from the BC FMNCP, how these outcomes are

223 achieved and whether these outcomes will be sustained after the program ends.

224

225 METHODS

226 Randomized controlled trial

227 Study design

228 Using a parallel group RCT, we will collect data at three time points: baseline (Time 1; 0 weeks, June

229 2019), immediately following the BC FMNCP (Time 2; 10-15 weeks, October 2019) and 16-weeks after

230 the BC FMNCP ends (Time 3; 26-31 weeks, February 2020).

231

232 Program overview

233 The BC FMNCP functions through a collaborative partnership between the BC Association of Farmers’

234 Markets, the BC Ministry of Health, farmers’ markets and community partners (i.e., local non-profit

235 organizations such as food banks, community services, and seniors’ societies). The BC Association of

236 Farmers’ Markets supports, develops and promotes farmers’ markets across BC [94] and oversees the

237 operations of the FMNCP. The FMNCP is supported by the province of BC and the Provincial Health

238 Services Authority. Community partners distribute coupons to program participants from their

239 organization locations and offer nutrition skill-building activities such as cooking classes or community

240 gardens to support intake of healthy foods purchased from farmers’ markets.[94]

241

Page 11 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



10

242 Recruitment

243 The FMNCP Director will identify approximately 15 BC communities for the study (from the existing

244 FMNCP and from program waiting lists) with the aim of achieving similar rural/urban coverage as the

245 existing FMNCP. Within each community, the FMNCP director will recruit community partners by

246 contacting those who have previously expressed interest in participating in the program, are members

247 of the BC Association of Farmers’ Markets, and offer nutrition skill-building activities for low-income

248 groups. Community partners within study communities will be responsible for identifying and enrolling

249 eligible low-income adults into the study. Community partners will recruit participants from among their

250 existing clients and will share study details with participants via phone, email or in-person, using posters

251 and other recruitment aids as needed. To minimize expectancy bias, whereby communication of

252 expected study outcomes influences participants’ behavior,[100-103] community partners will not

253 divulge specific details of the study’s objectives. Community partners will assess eligibility using a

254 screening questionnaire and will obtain voluntary, informed consent from eligible participants.

255

256 Patient and Public Involvement

257 All aspects of this study were co-designed with managers from the BC FMNCP who are directly involved

258 in delivering the program. They were engaged at the outset to ensure the study would answer questions

259 of importance to them. Although program participants did not directly participate in study design,

260 evidence pertaining to the life circumstances and challenges that low-income populations may

261 encounter was considered. Community partners across BC will recruit participants into the study and

262 will support study participants in completing study surveys. A study helpline will allow community

263 partners and participants to contact researchers for support throughout the study. We will report key

264 study findings to community partners and study participants, among other stakeholders, both in-person

265 (facilitated deliberative dialogue) and in written format (lay summaries).

Page 12 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



11

266

267 Eligibility criteria

268 Individuals will be eligible to participate if they meet the following criteria:

269 Adults (age ≥ 18 years)

270 Low-income as determined by community-specific thresholds (~$18,000/year annual household

271 income before taxes)

272 No expected change in household income prior to study completion

273 8 people living in the home (including the participant)

274 No expected change in household composition prior to study completion

275 Primary food shopper for the household

276 Does not have dementia or Alzheimer’s Disease

277 Able to speak, read and write in English (or have someone who can assist them)

278 No plans to move from principal residence prior to study completion

279 Has not previously participated in the BC FMNCP

280

281 Randomization

282 Following baseline data collection, eligible participants will be randomized to the FMNCP group (n=132)

283 or a no-intervention control group (n=132), with a 1:1 allocation ratio. An independent researcher from

284 the Clinical Research Unit at the University of Calgary will generate a blocked randomization sequence

285 that stratifies participants into blocks according to sex (male, female), geographic location (rural, urban),

286 pregnancy (yes, no) and breastfeeding (yes, no). Blocked randomization will help to ensure balanced

287 representation of participants in study arms.[104] REDCap (Research Electronic Data Capture), a secure,

288 web-based data collection and management application [105] hosted at the University of Calgary, will be

289 used to randomize participants into the FMNCP and control groups on the basis of this randomization

Page 13 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



12

290 sequence. The study coordinator will subsequently communicate participant group assignments to

291 community partners and participants. Allocation concealment will be ensured via secure storage of the

292 randomization sequence separately from the participant database, which will only be accessible by the

293 study coordinator and the Clinical Research Unit. Researchers will remain blinded to respondent

294 condition throughout the study. Although participants cannot be blinded to group allocation, they will

295 be blinded to the specific study objectives to reduce expectancy bias.

296

297 Intervention

298 In the existing BC FMNCP, community partners distribute one to two sheets of coupons per week (each

299 sheet contains $21 in coupons) to program participants for a total of 16 sheets. Coupons can be used

300 over 16-20 weeks to purchase fruits, vegetables, dairy, meat/poultry/fish, eggs, nuts, and cut herbs at

301 participating BC farmers’ markets. However, to allow sufficient time to recruit participants for this study,

302 community partners will distribute 16 coupon sheets to the FMNCP group over 10-15 weeks

303 (households with 5-8 individuals will receive 32 coupon sheets). To ensure participants receive all 16

304 coupon sheets, community partners will provide two coupon sheets per household during the first 1-6

305 weeks of the intervention. Participants in the FMNCP group will be invited to participate in nutrition

306 skill-building activities (e.g., cooking classes) offered by community partners throughout the

307 intervention period, however participation is not required (this is consistent with the existing FMNCP).

308 The types and frequency of nutrition skill-building activities offered vary across community partners. For

309 the duration of the study, the control group will not receive coupons nor be eligible to participate in

310 nutrition skill-building activities but will be eligible to participate in the BC FMNCP the following farmers’

311 market season. As participants in the control group already receive other supports from community

312 partners, they will continue to meet with their community partner as they normally would throughout

313 the intervention period.

Page 14 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



13

314

315 Data collection

316 Data will be collected from the FMNCP and control groups at three time points: Time 1: baseline (0

317 weeks), Time 2: immediately post-intervention (10-15 weeks), and Time 3: 16-weeks post-intervention

318 (26-31 weeks). At each time point, participants will first complete a questionnaire assessing

319 sociodemographic characteristics, health-related variables, and secondary and exploratory outcomes,

320 followed by a 24-hour dietary recall to assess diet quality (Table 1). The questionnaire and dietary recall

321 will be integrated within a web-based platform developed and pilot tested by the researchers [106]. A

322 second dietary recall will be completed 2-5 days later to better estimate usual intake and account for

323 within-individual variation in diet quality. All participants will receive cash incentives valued at $20 at

324 baseline and $40 at each Time 2 and 3. Participants will also receive small gift baskets containing useful

325 items donated from various BC organizations prior to data collection at Time 2 and 3, which will serve as

326 a reminder for the upcoming data collection.

327

328 At baseline, researchers will provide participants with a username and password to access the web-

329 based platform. Participants will be encouraged, but not required, to complete baseline data collection

330 at community partner locations immediately after providing informed consent. Community partners will

331 record whether data collection was completed on- or off-site. Immediately post-intervention and at 16-

332 weeks post-intervention, participants will receive an email requesting that they complete data collection

333 at a location of their choice (i.e., at home or at the community partner location). If data collection is not

334 completed within 48-hours of the initial prompt, researchers will make up to four additional attempts to

335 contact participants by email and/or phone.

336

337 Questionnaire

Page 15 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



14

338 The questionnaire will be administered via REDCap at all three time points to collect data on

339 sociodemographic characteristics, health-related variables, sense of community, mental well-being, food

340 insecurity, malnutrition risk, and subjective social status. Questions related to the FMNCP intervention

341 (e.g., coupon receipt) will be included in the questionnaire at Time 2 only.

342

343 Sociodemographic characteristics and health-related variables

344 Sociodemographic characteristics and health-related variables that will be assessed include date of

345 birth, sex, race/ethnicity, years lived in Canada, marital status, household size, number of children living

346 in the home, perceived physical health, pregnancy/breastfeeding, smoking status, height, weight,

347 educational level, employment status, annual household income, main source of income, and

348 community of residence.

349

350 Mental well-being

351 Mental well-being will be assessed using the valid 14-item Warwick-Edinburgh Mental Well-Being

352 Scale.[107] Scale items are positively phrased and assess various aspects of mental well-being such as

353 positive affect (e.g., feelings of optimism), psychological functioning (e.g., self-confidence, self-

354 acceptance) and satisfaction with interpersonal relationships.[108-110] The scale has been validated in a

355 variety of age, sex, and socioeconomic status groups [111] and cultural contexts [107] and has

356 demonstrated high test-retest reliability with an intra-class correlation of 0.83.[111] Responses are

357 scored on a 5-point Likert scale from 1 (none of the time) to 5 (all of the time) and are summed to

358 provide a single score ranging from 14 to 70.[107, 109] A higher score indicates higher perceived mental

359 well-being.[107]

360

361 Household food insecurity

Page 16 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



15

362 Household food insecurity will be assessed using Health Canada’s validated 18-item Household Food

363 Security Survey Module (HFSSM), which includes a 10-item adult scale and an 8-item child scale for

364 households with children under 18 years of age.[112] The HFSSM typically assesses experiences of

365 household food insecurity over the past year;[11] however, it will be modified to assess experiences of

366 household food insecurity in the past month. The HFSSM assesses experiences of marginal (one

367 affirmative response), moderate (adult subscale 2-5 affirmative responses or child subscale 2-4

368 affirmative responses) and severe (adult subscale ≥ 6 affirmative responses or child subscale ≥ 5

369 affirmative responses) food insecurity (18 items).[112] The HFSSM has been validated in a variety of

370 population groups and languages,[113, 114] and one study from the U.S. showed test-retest reliability

371 with a Pearson correlation coefficient of r=0.75.[115]

372

373 Sense of community

374 Sense of community will be assessed using the validated 8-item Brief Sense of Community Scale.[116]

375 Scale components are designed to assess each sense of community dimension according to the

376 McMillan-Chavis Model [117, 118] for sense of community, which includes four elements: membership,

377 influence, integration and fulfillment of needs, and a shared emotional connection.[117, 118] Each item

378 is scored using a Likert Scale of 1 (strongly disagree) to 5 (strongly agree).[116] Total sense of

379 community scores can range from 8 to 40 with a higher score indicating greater needs fulfillment, group

380 membership, influence and emotional connection within the community.

381

382 Risk of malnutrition

383 Risk of malnutrition will be calculated using the validated Malnutrition Universal Screening Tool

384 (MUST).[119, 120] The MUST assesses malnutrition risk using body mass index (BMI) (scored as 0 = BMI

385 > 20, 1 = BMI 18.5-20, 2 = BMI < 18.5), unplanned weight loss in the past 3-6 months (scored as 0 = < 5%

Page 17 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



16

386 of body weight, 1 = 5-10% of body weight, 2 = > 10% of body weight) and acute disease effect score

387 (acute illness with no or likely no nutritional intake for > 5 days).[120] Unplanned weight loss will be

388 modified to the past 3 months to accommodate the study timeline. In addition, acute disease effect

389 typically necessitates hospitalization and is unlikely to occur in community settings or clinics [120] and

390 will, therefore, be excluded.[121] Overall risk of malnutrition will be obtained by adding together

391 subscores for BMI and unplanned weight loss, with 0 indicating low risk, 1 indicating medium risk, and 2

392 or more indicating high risk of malnutrition.[121] The MUST has demonstrated a high test-retest

393 reliability with a Cohen’s kappa coefficient of =0.94.[122]

394

395 Subjective social status

396 Subjective social status will be assessed using the validated MacArthur Scale of Subjective Social Status

397 community ladder [123, 124] which consists of a single-item visual analog scale whereby respondents

398 place themselves on a ladder rung according to their perceived social standing relative to others in their

399 community.[123, 125] Responses can take a value from 1 to 10, with a higher score indicating higher

400 perceived social status.[123]

401

402 FMNCP intervention data

403 At Time 2 only, participants will report whether they received FMNCP coupons and attended nutrition

404 skill-building activities (to assess contamination of the control group), how often and how much of their

405 own money was spent at farmers’ markets during the intervention period and the types of foods

406 purchased.

407

408 Dietary intake

Page 18 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



17

409 Participants will complete two unannounced dietary recalls at each time point. Twenty-four hour dietary

410 recalls are a recommended dietary assessment method to evaluate the effect of an intervention on diet

411 quality, as they have less systematic error than other self-reported dietary assessment tools.[126, 127]

412 Administration of unannounced dietary recalls minimizes reactivity bias, where participants adjust their

413 dietary intake in anticipation of having to report it.[128]

414

415 Participants will record all foods and beverages consumed (excluding supplements) from midnight to

416 midnight the previous day using Health Canada’s validated Automated Self-Administered 24-hour

417 Dietary Recall (ASA24-Canada-2018),[127, 129-131] an automated online dietary assessment tool.[130,

418 131] The ASA24-Canada-2018 collects information regarding dietary intake in a series of four steps: 1)

419 foods consumed at each meal/snack, 2) queries regarding omitted meals/snacks if gaps between eating

420 occasions are 3 hours, 3) details of cooking methods, portions, and additions (e.g., condiments), and 4)

421 review of commonly forgotten items.[127, 132] The ASA24 concludes with a final question querying

422 whether reported intake was less than usual, usual, or more than usual.[132] The ASA24 has been used

423 with older, multi-ethnic, and disadvantaged adults [101-103, 127, 133] and was preferred by a majority

424 of participants compared to interviewer-administered recalls;[101] however, in a recent study among BC

425 FMNCP participants, we identified several usability issues with the ASA24.[106] For example,

426 participants reported difficulties in searching for specific foods and making changes once meals were

427 entered.[106] We will aim to address some of these challenges by including a brief pictorial user guide in

428 survey invitation emails and by training community partners to assist participants with the ASA24-

429 Canada-2018. In addition, participants and community partners will have access to a toll-free study help-

430 line available 10 hours/day, 6 days/week during data collection. Help-line operators include three

431 registered dietitians and the study coordinator, all of whom completed a half day group training session.

Page 19 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



18

432 Interrater reliability in entering meals into the ASA24-Canada-2018 among the four help-line operators

433 was high, with an intraclass correlation of 0.98.

434

435 The purpose of the help-line is two-fold: 1) to serve as a support platform for community partners to ask

436 questions, update researchers, and share feedback, and 2) to assist participants in completing data

437 collection and to answer any study-related questions. If needed, help-line operators will verbally read all

438 questions to participants and enter their responses online on their behalf. To maintain blinding,

439 operators will remind participants not to disclose their group assignment during the call. Supporting

440 participants during data collection will help to minimize missing and inaccurate data, and participant

441 attrition. Furthermore, providing a means through which community partners and participants can

442 contact researchers directly will allow researchers to build rapport with community partners and

443 participants.

444

445 Data collected by community partners and farmers’ market vendors

446 Community partners will maintain records of the number of coupons distributed to each participant

447 (coupons will have a unique bar code number that will be recorded for each participant) and the

448 frequency and types of nutrition skill-building activities each participant attended. Coupon redemption

449 and foods purchased with each coupon will also be tracked by farmers’ market vendors, who will note

450 foods purchased with each coupon (i.e., fruits, vegetables, dairy, meat/poultry/fish, eggs, nuts, cut

451 herbs) by using checkboxes on the back of each one. Farmers’ market managers will collect redeemed

452 coupons from vendors and submit them to the FMNCP. They will complete tracking sheets noting the

453 number of coupons redeemed and foods purchased with coupons.

454

Page 20 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



19

455 Table 1 Randomized controlled trial outcomes and measurement tools

Outcome Method Measurement toolPrimary outcomeOverall diet quality Two 24-hour dietary recalls at

Time 1, 2, and 31, 2, 3 Automated Self-Administered 24-

hour Dietary Recall Healthy Eating Index-2015

Secondary outcomesDiet quality subscores Two 24-hour dietary recalls at

Time 1, 2, and 31, 2, 3 Automated Self-Administered 24-

hour Dietary Recall Healthy Eating Index-2015

Sense of community Questionnaire at Time 1, 2, and 31, 2, 3

Brief Sense of Community Scale

Mental well-being Questionnaire at Time 1, 2, and 31, 2, 3

Warwick-Edinburgh Mental Well-being scale

Food insecurity Questionnaire at Time 1, 2, and 31, 2, 3

Household Food Security Survey Module

Malnutrition risk Questionnaire at Time 1, 2, and 31, 2, 3

Malnutrition Universal Screening Tool

Exploratory outcomeSubjective social status

Questionnaire at Time 1, 2, and 31, 2, 3

MacArthur Scale of Subjective Social Status community ladder

456 1Time 1: baseline (0 weeks)457 2Time 2: immediately post-intervention (10-15 weeks)458 3Time 3: 16-weeks post-intervention (26-31 weeks)459

460 Data Analysis

461 Healthy Eating Index-2015

462 Diet quality scores will be calculated using the validated Healthy Eating Index-2015 (HEI-2015),[134-138]

463 a tool used to assess conformance with the 2015-2020 Dietary Guidelines for Americans.[137] HEI scores

464 are associated with indicators of socioeconomic position [139] and chronic disease.[22, 136, 140-143]

465 Although Canadian adaptations of the HEI have been developed, they have either not been validated,

466 are not density-based, or reflect dietary recommendations that are no longer current.[13, 144] Given

Page 21 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



20

467 that dietary recommendations in Canada and the U.S. are similar,[145-147] the HEI-2015 remains an

468 appropriate tool to assess diet quality of Canadians.[13]

469

470 The HEI-2015 encompasses thirteen dietary components to assess overall diet quality,[137] including

471 nine ‘adequacy’ components (foods/nutrients recommended to consume, including total fruits , whole

472 fruits, total vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant

473 proteins, fatty acids) and four ‘moderation’ components (foods/nutrients recommended to limit,

474 including refined grains, sodium, added sugars, saturated fats). Component scores are density-based

475 and, therefore, independent of energy intake.[137, 148] Diet quality (total HEI-2015 scores and

476 subscores) will be calculated using the simple HEI scoring algorithm.[137] This method provides scores

477 at the individual level and can, therefore, accommodate the multi-level nature of our data and include

478 covariates.[137] HEI-2015 scores will be calculated using three nutrient databases linked to the ASA24-

479 Canada-2018:[149] the Canadian Nutrient File and the U.S. Department of Agriculture’s (USDA) Food

480 and Nutrient Database for Dietary Surveys to convert dietary intakes to energy and nutrient intakes, and

481 the USDA Food Patterns Equivalents Database to convert dietary intakes to dietary constituents (e.g.,

482 fruits) and measurement units consistent with HEI-2015 scoring standards (e.g., cup-equivalents of

483 fruit).[138, 148, 150, 151] Ratios for each of the dietary constituents (e.g., quantity of fruit per 1000

484 kcal) will be calculated for each participant and scored using HEI-2015 scoring standards. The total score

485 for each participant will be derived by adding the scores for intake of ‘adequacy’ and ‘moderation’

486 components with possible scores ranging from 0-100. A higher score indicates a higher quality diet and,

487 thus, greater conformance with dietary guidelines.[137]

488

489 Statistical analyses

Page 22 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



21

490 Descriptive analyses will be conducted to examine participant characteristics by group at each time

491 point. Characteristics of study completers (i.e., provided data at Time 3) and non-completers will also be

492 compared.

493

494 Analyses will be intention-to-treat, in which participants will be analysed within the groups to which

495 they were randomized regardless of adherence (e.g., failure to redeem coupons) or dropout. The

496 analyses will include all participants who provided data at baseline. Repeated measures mixed-effect

497 regression will assess differences in changes in mean HEI-2015 scores, HEI-2015 subscores, mental well-

498 being, sense of community, and subjective social status between the FMNCP and control groups

499 immediately post-intervention and 16-weeks post-intervention. Multinomial logistic regression will be

500 used to assess differences in the odds of experiencing household food insecurity and risk of malnutrition

501 for the FMNCP group compared to the control group immediately post-intervention and 16-weeks post-

502 intervention. Statistical models will include intervention group (FMNCP vs control), time from baseline,

503 intervention-by-time interaction, blocking variables (i.e., sex, rural/urban, pregnancy, breastfeeding),

504 baseline values of the outcome, questionnaire/dietary recall mode (online, phone), household size, and

505 place of data collection (community partner, other) as fixed effects covariates. Participant-specific (i.e.,

506 repeated measures) and rural/urban variations in outcomes will be modeled using random effects.

507 Models will also include covariates specific to each outcome to increase the precision of estimates.[87]

508 For the primary outcome of overall diet quality, models will include the following: children living in the

509 home (yes, no), sex, age, BMI, marital status, race/ethnicity, perceived health, smoking, day of data

510 collection, and dietary recall number (i.e., dietary recall 1 or 2). Models that are and are not adjusted for

511 an indicator of energy intake misreporting (the ratio of reported energy intake to total estimated energy

512 expenditure) will be presented. Adjusted group differences (i.e., FMNCP group vs control group) in

513 outcomes will be estimated using 95% confidence intervals and corresponding p-values.

Page 23 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



22

514

515 Subgroup analyses will examine whether the impact of the intervention on primary and secondary

516 outcomes differs according to age group or sex. Dose-response analyses will examine whether the

517 impact of the BC FMNCP on overall diet quality depends on the number of coupons redeemed and the

518 number of nutrition skill-building activities attended. Given that coupon redemption typically exceeds

519 90% [94], the former analysis may not be possible. Interactions will be retained in statistical models if

520 p


23

538 Sample size calculation

539 The sample size was calculated from a RCT that investigated the impact of a fruit and vegetable rebate

540 on HEI-2010 scores in low-income participants in the U.S.,[87] and a cross-sectional study that assessed

541 average diet quality scores in disadvantaged Canadians.[42] In the RCT, diet quality in the intervention

542 group was 4.7 points higher (95% CI 2.4, 7.1) at follow-up compared to controls.[87] This difference can

543 be translated to, for example, an additional half serving of fruit per day, which is clinically meaningful

544 and achievable.[97, 140] Assuming a type I error of 5%, an attrition rate of 30% by the 26- to 31-week

545 follow up, and potential design effects based on sampling within different communities (estimated at

546 1.1, or an inflation of 10%), 264 participants are required for 80% power to detect a 4.7-point difference

547 in diet quality.

548

549 Longitudinal qualitative investigation

550 A longitudinal qualitative study will be conducted concurrently with the RCT. Qualitative findings will

551 provide in-depth data pertaining to participants’ experiences in accessing nutritious foods, perceptions

552 of program outcomes and whether perceived outcomes are sustained once subsidies end.

553

554 Methodology and theoretical framework

555 Methodology

556 Qualitative description will be used as a methodological approach [157] to provide rich descriptions of

557 participant experiences of accessing nutritious foods, perceived short-term outcomes from the program

558 and how outcomes were achieved. The aim of qualitative description is to provide an accurate account

559 and description of events and to gain insight on how participants view and interpret those events.[157-

560 159] Sullivan-Bolyai et al [160] suggest that by providing rich description of participants’ experiences,

561 findings from qualitative description are useful for informing and refining interventions that seek to

Page 25 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



24

562 address health inequities. Thus, this methodology will allow for an in-depth exploration of BC FMNCP

563 participants’ experiences and perceptions related to the program.

564

565 Theoretical framework

566 Data generation and analysis will be guided by Freedman et al’s [161] theoretical framework of

567 nutritious food access. The framework was developed to inform programs and policies to increase

568 nutritious food access within economically disadvantaged groups and was developed from data

569 generated by low-income shoppers in farmers’ markets.[161] The model includes five domains: 1)

570 economic (e.g., household finances, incentives), 2) spatial-temporal (e.g., location, transportation), 3)

571 service delivery (e.g., ease of coupon redemption), 4) social (e.g., social interaction, stigma), and 5)

572 personal factors (e.g., nutrition knowledge, health status). The theoretical framework of nutritious food

573 access highlights economic factors as key determinants of nutritious food access, while emphasizing the

574 need for multidimensional and multi-level policies and interventions that improve food access for low-

575 income households.[161]

576

577 Sampling and recruitment

578 The sampling procedure will include collaboration with three community partners from two rural and

579 one urban community that are part of the BC FMNCP, but that are not participating in the RCT.

580 Individuals participating in the RCT will not be eligible to participate, as interviews could prompt

581 additional behavior change or differential reporting for the RCT. Within the selected communities, 20-25

582 adults from low-income households currently enrolled in the existing BC FMNCP will be purposefully

583 selected to be representative of FMNCP participants. Community partners will approach individuals

584 participating in the existing BC FMNCP and will conduct preliminary screenings for eligibility and obtain

585 informed consent. Screening and signed informed consent forms will be shared with researchers prior to

Page 26 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



25

586 beginning interviews. The adequacy of the sample will be continuously re-evaluated throughout the

587 data generation and analysis process.[162] In recruiting a sample that is representative of BC FMNCP

588 participants, we will target mainly adults with children and older adults who are receiving coupons for

589 the first time. Additional eligibility criteria include individuals who are the primary food shopper for the

590 household, can communicate in English, have 8 or fewer people living in the home (including the

591 participant), are not planning to move from their principal residence nor expecting any major changes to

592 their annual household income prior to the second interview, and are willing to participate in two

593 interviews. Participants will be offered a $25 and $35 cash incentive following the first and second

594 interview, respectively, and a small gift basket between time points.

595

596 Data generation

597 To reach study objectives, data generation will occur at two time points: between weeks 8 and 12 of the

598 BC FMNCP and 8-10 weeks after the program ends. Semi-structured individual interviews will be

599 conducted to better understand individual experiences and perceptions related to the BC FMNCP.

600 Interviews will be conducted by two researchers with previous qualitative research experience. These

601 researchers will develop an initial semi-structured interview guide for the first time point, guided by the

602 five domains of Freedman et al’s [161] theoretical framework. The initial interview guide will be

603 designed to capture individual experiences of participating in the BC FMNCP and perceived outcomes of

604 the program, as well as how these outcomes were achieved. Participants will be asked open-ended

605 questions such as ‘What have been your experiences with the Farmers’ Market Nutrition Coupon

606 Program so far?’ and ‘Why do you go to farmers’ markets?’, followed by more specific questions

607 capturing facilitators and barriers of accessing nutritious foods and perceived outcomes of the program.

608 Follow-up interviews will also be semi-structured and will explore themes that were identified from the

609 first round of interviews in order to ascertain and examine differences, similarities and changes in

Page 27 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



26

610 participant experiences of the program and to examine the sustainability of perceived program

611 outcomes once the program ends. These themes will be explored cross-sectionally by asking participants

612 about their current experiences and perceived outcomes of no longer receiving subsidies, as well as

613 longitudinally by exploring specific changes in relation to themes, and why and how those changes

614 occurred. Additional themes related to participants’ experiences with the BC FMNCP may also be

615 developed through further analysis and explored with follow-up questions.

616

617 In-person, 60-minute semi-structured interviews will be conducted with participants at both time points.

618 Following each interview, basic demographic information will be collected such as sex, age,

619 race/ethnicity, marital status, household composition, household income, education, employment

620 status and household food insecurity. Interviews will be audio recorded and transcribed verbatim. The

621 two researchers will interview the same individuals at each time point to enhance consistency and

622 rapport with participants. Descriptive field notes will also be recorded during each interview to capture

623 information on the setting and respondents’ reactions to questions.

624

625 Data analysis

626 Data analysis will be iterative and conducted by both researchers in two phases, including a cross-

627 sectional analysis at each time point and a longitudinal analysis to identify similarities, differences and

628 changes over time. NVivo software (version 12.3, QSR International Pty Ltd) will be used to manage and

629 organize coding. In the first phase of the analysis, data will be analyzed separately at each time point.

630 The two researchers will use directed content analysis [163] to analyze the data, using the five domains

631 of Freedman et al’s [161] framework to guide development of an initial coding scheme. The analysis will

632 be semi-deductive based on the theoretical structure of the framework. Data analysis will begin with

633 repeated listenings of audio files and reading and re-reading of corresponding transcripts and writing

Page 28 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



27

634 down analytic notes related to Freedman et al’s five domains. Data that do not fit within the

635 framework’s five domains will be coded inductively.[164] Following the first four interviews at each time

636 point, the two researchers will independently analyze interviews and will subsequently meet to reach

637 consensus about a coding scheme that captures the major ideas identified. The two researchers will

638 then conduct four more interviews, after which they will calculate agreement of coding practices based

639 on the coding scheme, aiming for a threshold of 80% as the criterion of acceptability.[161] Ongoing

640 meetings between the two researchers will permit iterative adjustments to the interview questions and

641 coding scheme as data generation and analysis proceed. The two researchers will then use the

642 interpretive practice of constant comparison and memoing to ensure that interpretations are consistent

643 with the underlying data and that relationships between codes are clear and consistent,[163] after

644 which they will collate and categorize codes to generate themes.[163, 165] The second set of interviews

645 will allow for further exploration of themes identified from the initial interviews, along with new themes

646 that may be developed from no longer participating in the program. Analysis will be conducted by the

647 two researchers and will be guided by the coding scheme from the first set of interviews. The coding

648 scheme will be revised as needed to reflect data from the second set of interviews. Once cross-sectional

649 themes have been developed for each time point, they will be presented to the research team for

650 review.

651

652 Following initial data analysis at both time points, a longitudinal analysis will be conducted in which the

653 data will be integrated to explore changes, similarities and differences between time points. For this

654 analysis, one researcher will examine cross-sectional themes from both time points simultaneously,

655 guided by Saldana’s [166] descriptive and analytic/interpretive questions. Answers to these questions

656 are iterative and may arise at any point during the analysis.[166] From this process, findings will be

657 integrated into longitudinal themes that focus on participants’ experiences, including facilitators and

Page 29 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



28

658 barriers of accessing nutritious foods during and after the BC FMCNP, perceived program outcomes and

659 how they were achieved, and whether and how perceived outcomes were sustained over time.

660

661 Strategies to enhance rigor

662 Potential limitations and threats to the trustworthiness of study findings will be offset by applying

663 strategies suggested by Noble and Smith [167] to enhance rigor, including:

664 1) Truth value. For a fuller representation of participants’ accounts and perceptions, peer debriefing

665 between the two researchers will support reflexivity and assist researchers to uncover assumptions

666 throughout data generation and analysis. The use of thick descriptions from participant interviews will

667 help ensure that findings remain true to participants’ accounts.[167, 168]

668 2) Consistency and confirmability. The researchers will create an audit trail to provide a transparent

669 description of study processes,[169, 170] which will include a log of decisions made throughout the

670 study, personal reflections to enhance transparency of the research process, raw data from interviews,

671 field notes during data generation, and analysis products such as summaries and notes.[167] Many of

672 these details will also be included in study manuscripts.

673 3) Applicability. Researchers will provide rich details of the study context and thought processes during

674 data generation, analysis and interpretation in published findings. This will allow readers to assess the

675 transferability of study findings to other contexts or settings.

676

677 ETHICS AND DISSEMINATION

678 Ethics approval was obtained from the Conjoint Health Research Ethics Board of the University of

679 Calgary (REB18-0508) (Calgary, Alberta, Canada), University Ethics and Compliance from Rutgers

680 University (FWA00003913) (Piscataway, New Jersey, U.S.), and the Office of Research Ethics from the

681 University of Waterloo (ORE #40724) (Waterloo, Ontario, Canada). Ethics boards, researchers,

Page 30 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



29

682 community partners, the BC FMNCP manager, and participants will be informed of any study protocol

683 modifications (e.g., eligibility criteria, study procedures) that impact the conduct of the study. Reporting

684 will adhere to CONSORT, TIDieR, and SPIRIT reporting standards. Study findings will be presented to

685 stakeholders within government and communities across Canada to inform decision-making via a

686 facilitated deliberative dialogue, policy briefs and lay summaries. Results will also be disseminated

687 through peer-reviewed journal publications and conference presentations.

688

689 STUDY MANAGEMENT AND MONITORING

690 The principal investigator (DLO) will manage and oversee the study, review the study protocol, and assist

691 with organising study committees and meetings. All researchers are considered steering committee

692 members, responsible for reviewing and agreeing on protocol modifications, as needed.

693

694 DATA MANAGEMENT

695 Participants will be issued a unique study identification number for all data collected. All personally

696 identifying information will be kept separate from study data and stored securely on password

697 protected computers. Survey data will be stored on both the ASA24 and REDCap servers, which are

698 secure and managed by their respective organizations. All data downloaded from these systems will be

699 de-identified and only the researchers will have access to the data. A formal data monitoring committee

700 will not be established as study-associated risk are minimal. Researchers will review all data within 24

701 hours of receipt and will contact participants regarding missing or implausible data. No interim analyses

702 or stopping guidelines have been established.

703

704 DISCUSSION

705 Inequities in diet-related chronic disease are an ongoing public health concern.[171-173] Given that the

Page 31 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



30

706 determinants of dietary patterns are complex and multifactorial,[18, 42, 48] it is crucial that public

707 health initiatives address all socioecological levels to reduce dietary and health inequities in low-income

708 populations.[54, 55, 174] The BC FMNCP has the potential to improve diet quality, health, and

709 psychosocial well-being of low-income participants, as it is a multi-component program that links the

710 agricultural and health sectors and addresses determinants of health and dietary intake at all levels of

711 the socioecological model, including individual (i.e., nutrition skill building activities), social (e.g.,

712 interpersonal interactions in farmers’ markets), community (e.g., improved access to healthy foods), and

713 policy levels (e.g., the program offers government-funded food subsidies).

714

715 Previous studies that have assessed outcomes from farmers’ market food subsidy programs are often

716 limited by weak study designs, short follow-up times, use of brief fruit and vegetables screeners, and

717 primarily examined fruit and vegetable intake rather than overall diet quality.[67] This study will take

718 steps toward filling these gaps by using a RCT design that will provide the first evidence of the causal

719 impact of a farmers’ market healthy food subsidy on the overall diet quality (primary outcome), diet

720 quality subscores, mental well-being, sense of community, experiences of food insecurity, risk of

721 malnutrition (secondary outcomes), and subjective social status (exploratory outcome) of low-income

722 adults, and will provide evidence of whether changes in outcomes among program participants are

723 maintained over time. This study will also assess dietary intake using 24-hour dietary recalls, which have

724 less systematic error than short screeners or food frequency questionnaires [128] and will investigate

725 the impacts of a broad food subsidy, in contrast to previous studies that have examined solely fruit and

726 vegetable subsidies.[63, 66, 175] We will use valid measurement tools to increase accuracy of effect

727 estimates. Finally, the longitudinal qualitative evaluation is uniquely designed to explore participants’

728 experiences in, and perceptions of, accessing nutritious foods, including facilitators and barriers during

729 and after the BC FMNCP, and perceived short-term program outcomes and how these outcomes were

Page 32 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


jopen-2019-035143 on 5 May 2020. D

ownloaded from



31

730 achieved. A longitudinal approach will provide insight into whether and how participants’ experiences

731 change once the program ends and whether perceived program outcomes are sustained over time.

732 These data will complement findings from the RCT by providing rich descriptions of participant-

733 perceived program outcomes and can further explore perceived mechanisms of program impacts.[176,

734 177] In addition, these data will highlight changes in lived experiences over time, including causes and

735 processes of change that may occur due to no longer participating in the program.[178]

736

737 Study limitations

738 Alongside these important strengths, this study has methodological limitations. First, the data collected

739 through the questionnaire and 24-hour dietary recall are self-reported and, therefore, subject to self-

740 reported measurement bias,[126] including reactivity and social desirability biases.[126, 128] This study

741 will aim to minimize these biases by using self-administered online tools, which may reduce social-

742 desirability bias compared to interviewer administered surveys and recalls.[128] Although 24-hour

743 dietary recalls will be unannounced at baseline, dietary recalls immediately post-intervention and 16-

744 weeks post intervention will be less so, as participants will receive emails inviting them to complete data

745 collection and may take up to 48 hours to do so. Second, lower socioeconomic position may be

746 associated with lower computer literacy.[179, 180] Given that the surveys will be delivered via an online

747 platform, participants may experience difficulty completing the surveys, which may result in implausible

748 or missing responses. However, other evidence suggests that most low-income individuals have access

749 to and regularly use computers and the internet.[133, 181] Moreover, participants may complete data

750 collection at all three time points at a community partner location, where community partners can assist

751 them with data collection, and researchers will be available for assistance via the study help-line. As the

752 study is longitudinal, there is also a risk of loss to follow-up. To maximize retention, the study team will

753 work closely with community partners to maintain communication with participants between time

Page 33 of 55


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


http://bmjopen.bm

j.com/

BM


bmj open | a leading multidisciplinary medical journal - bmj ......distributed as the published...

Documents