Phuong H Nguyen, 1 * Sunny S Kim, 2 Sarah C Keithly, 3 Nemat Hajeebhoy, 4 Lan M Tran, 4 Marie T Ruel, 2 Rahul Rawat 2 and Purnima Menon 5

Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine ß The Author 2013; all rights reserved. Advance Access publication 14 November 2013 Health Policy and Planning 2014;29:1008 1020 doi:10.1093/heapol/czt083 Incorporating elements of social franchising in government health services improves the quality of infant and young child feeding counselling services at commune health centres in Vietnam Phuong H Nguyen, 1 * Sunny S Kim, 2 Sarah C Keithly, 3 Nemat Hajeebhoy, 4 Lan M Tran, 4 Marie T Ruel, 2 Rahul Rawat 2 and Purnima Menon 5 1 International Food Policy Research Institute, Hanoi, Vietnam, 2 International Food Policy Research Institute, Washington, DC 20006, USA, 3 Institute of Social and Medical Studies, Hanoi, Vietnam, 4 FHI360, Hanoi, Vietnam and 5 International Food Policy Research Institute, New Delhi 110012, India *Corresponding author. Phuong Hong Nguyen, International Food Policy Research Institute, Room 203-E4B Trung Tu Diplomatic Compound, No 6 Dang Van Ngu, Hanoi, Vietnam. E-mail: P.H.Nguyen@cgiar.org Accepted 8 October 2013 Background Although social franchising has been shown to enhance the quality of reproductive health services in developing countries, its effect on nutrition services remains unexamined. This study assessed the effects of incorporating elements of social franchising on shaping the quality of infant and young child feeding (IYCF) counselling facilities and services in Vietnam. Methods Results Conclusions Process-related data collected 12 months after the launch of the first franchises were used to compare randomly assigned Alive & Thrive-supported health facilities (AT-F, n ¼ 20) with standard facilities (SF, n ¼ 12) across three dimensions of service quality: structure, process and outcome that capture the quality of facilities, service delivery, and client perceptions and use, respectively. Data collection included facility assessments (n ¼ 32), staff surveys (n ¼ 96), counselling observations (n ¼ 137), client exit interviews (n ¼ 137) and in-depth interviews with mothers (n ¼ 48). Structure: AT-F were more likely to have an unshared, well-equipped room for nutrition counselling than SF (65.0% vs 10.0%). Process: Compared with SF providers, AT-F staff had better IYCF knowledge (mean score 9.9 vs 8.8, range 0 11 for breastfeeding; mean score 3.6 vs 3.2, range 0 4 for complementary feeding). AT-F providers also demonstrated significantly better interpersonal communication skills (score 9.6 vs 5.1, range 0 13) and offered more comprehensive counselling sessions. Outcome: Overall utilization of franchises was low (10%). A higher proportion of pregnant women utilized franchise services (48.9%), compared with mothers with children 6 23.9 months (1.4%). There was no quantitative difference in client satisfaction with counselling services between AT-F and SF, but franchise users praised the AT-F for problem solving related to child feeding. Incorporating elements of social franchising significantly enhances the quality of IYCF counselling services within government primary healthcare facilities, particularly their structural and process attributes. Provided that service 1008

SOCIAL FRANCHISING IMPROVES NUTRITION COUNSELLING IN VIETNAM 1009 utilization is improved through demand generation, this model has the potential to impact IYCF practices and child nutrition. Keywords Social franchise, nutrition counselling, infant and young child feeding, quality of care, Vietnam KEY MESSAGES Social franchising is being used increasingly to improve the quality, availability and utilization of health services worldwide, particularly for reproductive health in developing countries. This is the first study assessing the impact of social franchising on shaping the quality of infant and young child feeding (IYCF) counselling. Incorporating elements of social franchising into the existing healthcare system significantly enhances the structural (e.g. material and human resources) and process (e.g. capacities and service delivery) attributes of IYCF counselling service quality. However, enhanced quality of care also needs to include improved outcomes such as high rates of service utilization, which will require greater efforts in demand creation. Social franchising offers a feasible and sustainable model with demonstrated potential to significantly improve the quality of IYCF counselling services. Introduction Adequate infant and young child feeding (IYCF) practices contribute to healthy child growth and development. In Vietnam, however, IYCF practices are suboptimal and contribute to the country s high rates of childhood stunting (27.5%) and underweight (16.8%) (Hop et al. 2000; NIN 2012). Exclusive breastfeeding (EBF) up to 6 months of age is very low at 20%, with 16 out of 63 provinces reporting EBF rates of 1% or less, whereas bottle feeding is high at 35% (Le et al. 2011). Optimal complementary feeding (CF) practices are disrupted by early introduction of complementary foods (62%) (Le et al. 2011) and poor nutrient quality (Duong et al. 2005). It is well documented that child feeding practices improve when women receive high quality, timely counselling and support (Bhutta et al. 2008; Renfrew et al. 2012). Community-based initiatives using peer counsellors for EBF support have been shown to be feasible and successful in several countries (Anderson et al. 2005; Nankunda et al. 2006; Leite et al. 2005; Tylleskar et al. 2011). The use of counselling cards or job aids for IYCF counselling has also been shown to significantly improve the quality of counselling sessions (Katepa-Bwalya et al. 2011). In Peru, nutrition education interventions delivered through government health centres successfully improved child dietary intakes and growth (Robert et al. 2007). Unfortunately, integrated, high quality IYCF-focused counselling services needed to bring about change to child feeding behaviours are not readily available in Vietnam (Nguyen et al. 2011). Despite the supportive policy environment for improving child nutrition in Vietnam (Nguyen et al. 2011), the implementation of appropriate interventions remains problematic. Commune health centres (CHC), the primary access point for health care and the principal implementers of government nutrition programmes, have not yet systematically integrated IYCF counselling services into routine health care (A&T 2012; Lapping et al. 2012). Many health providers report that national guidelines are rich in information about the benefits of breastfeeding (BF) but lack practical guidance on how to provide counselling and support to mothers (A&T 2009). In addition, the quality, frequency and follow-up of in-service training on IYCF are inadequate (Almroth et al. 2008; A&T 2009). These problems are compounded by the limited resources available to local health facilities. Together, these constraints lead to limited availability of IYCF counselling services, or when available, the services are of low quality, lack message consistency and fail to address caregivers specific concerns about IYCF (A&T 2009). Social franchising is being used increasingly to enhance health services worldwide, particularly in developing countries (Montagu 2002). Social franchising applies commercial franchising concepts so that a brand identity is equated with quality services that help achieve social and health benefits. This is enabled primarily through standardized operating procedures, support and training. Most experiences in social franchising are from reproductive health and show overall improvements in service quality (Bishai et al. 2008), access to and utilization of services (Stephenson et al. 2004; Ngo et al. 2010), and client satisfaction (Agha et al. 2007a; Ngo et al. 2009a). In Vietnam, social franchising has been used successfully to enhance the quality of reproductive health and family planning services (Ngo et al. 2009c). Alive & Thrive (A&T), a 6-year initiative to reduce undernutrition and death caused by suboptimal IYCF practices, built on this experience and applied social franchising principles to improve access to good quality IYCF counselling services in Vietnam (Baker et al. 2013). This study aimed to determine the effects of social franchising of IYCF counselling services on the quality of care, by comparing A&T-supported health facilities (AT-F) to standard facilities (SF) across three dimensions of service quality structure, process and outcome. These dimensions examine the quality of facilities, service delivery, as well as client perceptions and use of services.

1010 HEALTH POLICY AND PLANNING Background of the M3t Trhi Bé ThoP franchise model A&T applied a social franchising approach to develop a model under the brand name M3t Trhi BéThoP (MTBT) ( The Little Sun in English). The model has two main purposes: (1) to standardize and monitor services to ensure that counselling on IYCF is uniform and of good quality; and (2) to build on the existing healthcare infrastructure and decentralized services to ensure utilization and sustainability (Baker et al. 2013). MTBT focused on good quality and relevant training to build the capacity of health workers who are in the position to encourage and support mothers to practise optimal IYCF. A&T established a partnership with the National Institute of Nutrition (NIN), which is responsible for Vietnam s National Nutrition Program, to act jointly as the national franchisor. A&T and NIN provide general oversight and authority, whereas provincial health authorities serve as sub-franchisors to coordinate franchise operations. Individual health facilities at the province, district and commune levels serve as franchisees that deliver the standardized IYCF counselling services. Health facilities were first assessed, and those that met the franchise selection criteria (e.g. facility capacity and provider qualifications) were invited to join the franchise network, provided that they adhere to requirements. The organizational structure of the MTBT franchise model and the criteria for franchise selection are provided in the Supplementary Material. Following a feasibility study and based on formative research, the franchisor developed the MTBT service package and operational guidelines which were integrated into a network of existing government health facilities, adhering to the franchise requirements. The franchisor and sub-franchisors worked together to upgrade franchisee health facilities, provide in-service training and materials necessary for service provision, certify franchisees to operate under the franchise brand (see Supplementary Material for certification criteria), and provide ongoing support and routine monitoring of performance. In turn, franchisees were expected to deliver the MTBT service package in accordance with the franchise standards. The MTBT franchise was designed to operate at various levels and types of health facilities. The most comprehensive package of services is offered at the commune level through CHC and includes (1) EBF promotion during the third trimester of pregnancy, (2) EBF support at time of delivery, (3) EBF management within 6 months postpartum, (4) CF education when infants are 4 5.9 months of age and (5) CF management when children are 6 24 months of age. These services are delivered through individual and group counselling sessions. In order to generate use of and demand for franchise services, invitation cards are delivered to targeted mothers by village health workers or local nutrition collaborators, promotional materials are disseminated, and various events are held. A mass media campaign to promote optimal IYCF practices was also launched 6 months after initiation of the franchise model. At the time of this study, the MTBT franchise was introduced in 782 government health centres, including 660 CHC. One common element of a social franchise that has not yet been incorporated into the MTBT franchise is a cost recovery system. There is currently no pricing standard for service fees, and MTBT franchise services are provided free of charge. The franchisor plans to gradually incorporate fee collection after assuring service quality and demand. Methods Study setting and context This study is part of a larger process evaluation of A&T s IYCF interventions conducted in four provinces (Thai Nguyen, Thanh Hoa, Quang Ngai and Vinh Long) of Vietnam (Rawat et al. 2013), within the framework of a cluster-randomized impact evaluation of the programme (Menon et al. 2013). These provinces span the northern, central and southern regions of the country and are geographically representative of the 15 provinces where A&T is in operation. Process-related data were collected between April and July 2012, about 12 months after the first launch of the MTBT franchise in AT-F. Standard facilities (SF) government healthcare facilities providing the standard level of care served as matched controls. Operational definition of quality of care Following Donabedian s conceptual framework for assessing quality of care, this study examined three components of service quality: structure, process and outcome (Donabedian 1988). Each of these components was measured by multiple indicators, as presented in Figure 1. Indicators of structure include the physical condition of the facility, materials, and supplies; availability and qualifications of personnel; and service availability. Process measures are related to the provision of services based on the capacity and performance of personnel, in accordance with the standard guidelines of the franchise. Outcome measures include service utilization and client satisfaction, including the reported likelihood of returning and recommending the service to others. Sampling and data collection Twenty AT-F and 12 SF from the same districts were randomly selected for this study. Nearly all of the MTBT franchise facilities fulfilled the minimum franchise criteria; the 20 AT-F were among them. Various methods were used to gather data on the three dimensions of service quality described earlier, including facility assessment (n ¼ 32), structured observation of counselling services (n ¼ 137), structured interview with health providers (n ¼ 96) and clients (n ¼ 137), and in-depth interview with franchise users and non-users (n ¼ 48). A summary of the topics covered and the sample type and size for each data collection method is provided in Table 1. The franchise certification criteria and other franchise materials were used to develop observation checklists, tools and questionnaires. Interviews and observations were conducted by public health researchers, who were trained and standardized for data collection with high inter-rater reliability. Counselling observations were opportunistic and conducted on the same days as the client exit interviews; one or more counselling sessions were led by the same health providers. Various indicators obtained from across data sources were used to determine the differences in structural attributes of service quality between AT-F and SF. Baseline data collected as part of the programme impact evaluation (Nguyen et al. 2010) were also used to assess any comparable characteristics of AT-F and SF between the baseline and process evaluation periods.

SOCIAL FRANCHISING IMPROVES NUTRITION COUNSELLING IN VIETNAM 1011 Figure 1 Conceptual framework for assessing quality of care. Table 1 Data collection methods and sample Methods Data collection topics Sample type and size Facility assessment Franchise set up and adherence to standards within health facilities (e.g. infrastructure, equipment, BCC materials), and utilization and coverage of IYCF-related services (using data from monitoring forms) 20 AT-F and 12 SF Health provider survey Counselling observation Exposure to training, IYCF and nutrition knowledge, motivation and satisfaction, supervision, interactions with other health providers, incentives and their role, and time commitments to IYCF counselling Health staff competence and performance during IYCF counselling, including interpersonal communication and technical skills 96 providers (60 from AT-F and 36 from SF) 137 counselling observations (101 at AT-F and 36 at SF) Client exit interview Experience and satisfaction of counselling service use 137 mothers with children 0 24 months who received IYCF counselling (101 at AT-F and 36 at SF) In-depth interview with mother Perceptions of and satisfaction with AT-F services, and the reasons for not using AT-F services 48 mothers with children 0 12 months in 4 A&T communes (27 franchise users and 21 non-users) Data analysis Quantitative data were analysed using Stata 11 (StataCorp 2009). Descriptive statistics were used to describe the sample characteristics of AT-F and SF, health providers and clients. To measure the association between facility type (AT-F or SF) and quality indicators, two-tailed parametric and non-parametric tests (P < 0.05) were applied. For process-related quality indicators, un-weighted summary scores were constructed from several items that constitute each indicator type. For example, health providers were asked various questions about BF and CF to assess their IYCF knowledge, and each correct answer received a score of 1. A simple additive score for the knowledge type (BF or CF) was created, summing each correct response. Similar summary scores were constructed for interpersonal communication skills and technical content provided during counselling sessions. The mean percentage of each correct item/ response and the overall mean scores for indicator types are presented in the results. Results of all service quality components are compared between AT-F and SF at process evaluation, as well as with any differences in change from baseline, wherever similar measures are available. Content analysis of the in-depth interviews was conducted using the qualitative data analysis software NVivo 10 (QSR International 2012). Interview transcripts were coded and

1012 HEALTH POLICY AND PLANNING analysed, applying a list of a priori thematic codes based on the research objectives, MTBT service guidelines and the interview guides. Outputs of code queries were interpreted and summarized to extract findings related to the service quality components. The study protocol was approved by the Institutional Review Boards of the Institute of Social and Medicine Studies in Vietnam and of the International Food Policy Research Institute. Results Sample characteristics Table 2 presents the characteristics of the CHC, health providers and clients who received IYCF counselling. Each CHC serves an average population of 7000 people in nine villages. AT-F (n ¼ 20) and SF (n ¼ 12) did not significantly differ in the types and numbers of population they serve or in the provider Table 2 Characteristics of study samples Characteristic A&T-supported facility Standard facility Per cent/mean SD Per cent/mean SD Profile of communes where CHC operate n ¼ 20 n ¼ 12 Number of villages/hamlets 8.4 4.2 9.0 4.0 Estimated population (thousand) 7351.2 2450.3 6812.6 3184.5 Number of pregnant women 67.5 32.0 56.4 28.5 Number of pregnant women in the third trimester 19.5 8.7 16.2 10.2 Number of deliveries in past year 100.4 38.4 90.5 40.4 Number of children < 5 years 520.2 188.1 441.1 200.8 Number of children < 2 years 205.1 71.3 171.4 72.4 Number of infants < 6 mo 44.8 23.3 40.9 21.9 Distance from CHC to district centre (km) 10.4 7.2 9.2 4.9 Profile of providers n ¼ 60 n ¼ 36 Age (y) 38.9 6.2 41.4 6.2 Gender (female) 80.0 69.4 Length of time living in the commune (y) 21.1 15.8 25.8 16.7 Education Secondary school 0.0 2.8 High school 6.7 8.3 College or higher 93.3 88.9 Position at CHC Physician 25.0 22.2 Physician assistant 33.3 44.4 Nurse/midwife 41.7 33.3 Profile of clients n ¼ 101 n ¼ 36 Client type Pregnant woman in third trimester 29.7 33.3 Mother of child < 6 mo 40.6 33.3 Mother of child 6 24 mo 28.7 33.3 Age 28.2 5.0 30.1 5.1 Ethnicity (Kinh) 93.1 91.7 Education Primary school or less 10.9 8.3 Secondary school 23.8 44.4 High school 34.6 33.3 College or higher 30.7 13.9 Occupation Farmer 35.6 41.7 Salary employee 30.7 16.7 Small trader/self-employment 18.8 27.8 Working from home/housewife 14.9 13.9 SD¼ standard deviation.

SOCIAL FRANCHISING IMPROVES NUTRITION COUNSELLING IN VIETNAM 1013 characteristics. Although this study was carried out 12 months after the launch of the first MTBT franchises, AT-F varied in their duration of operations (4 with <3 months, 4 with 3 6 months, 8 with 6.1 9 months and 4 with >9 months). The mean duration of franchise operation was 6.4 months. As intended by the study design, the sample of clients for exit interviews was equally distributed among different target populations (pregnant women in the third trimester and mothers of children under 6 months of age and aged 6 24 months). Less than 10% of the clients were of non-kinh ethnicity. Clients education and occupation did not vary significantly by facility type. Among the 48 in-depth interview respondents (results not shown), 48% were mothers of children under 6 months of age, and 52% were mothers of children aged 6 12 months. The average age was 28 years, and the main occupations were farmers, housewives or selfemployed/small traders. Structural attributes Infrastructure and equipment AT-F had significantly higher levels of quality infrastructure and equipment than those of SF, particularly in terms of the facility condition and types and quantity of equipment and materials available (Table 3). This was an expected finding because physical upgrades and material inputs were provided as part of the MTBT franchise setup. Although 65% of AT-F had a room dedicated solely to nutrition counselling, only 10% SF had a similar facility. Observation results indicated that the AT-F counselling areas were more colourful and attractive in appearance than those in SF. Also, AT-F offered amenities that were not available in SF such as play areas and toys for children. From the 14 potential equipment and supply items (e.g. table and chairs) to support IYCF counselling activities, AT-F had a significantly higher number of items available than SF (8.7 vs 3.9 items, respectively). AT-F were also more equipped with instruments for measuring anthropometry, kitchen utensils for cooking demonstrations and materials for behaviour change and communication (BCC). Availability of services and providers A broad range of nutrition-related services was offered at the CHC, including IYCF counselling and support for pregnant women and mothers, food demonstrations and growth monitoring. There was no significant difference in these general types of services offered by facility type (Table 3). A CHC was usually staffed by one doctor, who was supported by three to five other health staff such as physician assistants and nurses/ midwives. Again, there was no significant difference in availability of health providers by facility type, except for the significantly less community-based workers in AT-F (n ¼ 9.4) compared with SF (n ¼ 11.5) (Table 3). In-service training AT-F providers received more training in IYCF than SF providers (Table 2). Compared with only 13.9% of SF providers, 86.7% of AT-F providers reported having participated in an IYCF training course in the past 2 years. Also, training received by AT-F staff covered a statistically significantly higher number of key IYCF topics (13.2 for AT-F vs 5.8 for SF). Process attributes Provider capacity Providers knowledge on specific aspects of BF and CF and the summary scores of correct responses across knowledge types are presented in Table 4. BF knowledge was relatively high among both AT-F and SF staff, but AT-F providers had a statistically significantly higher overall score on BF knowledge (9.9 vs 8.8 mean score, range 0 11). AT-F providers also had statistically significantly greater knowledge on several individual BF topics: sufficiency of breast milk (33.3% vs 11.1%), not giving formula (96.7% vs 77.8%) or water (95.0% vs 58.3%) before the age of 6 months and breast milk expression (96.7% vs 77.8%). Also, AT- F providers recalled more BF benefits for children and mothers than SF providers. Knowledge of optimal CF practices was also statistically significantly higher among staff in AT-F than in SF (3.6 vs 3.2 mean score, range 0 4), mainly driven by the higher knowledge of iron-rich foods (86.7% vs 69.4%). Other specific CF topics did not vary significantly by facility type. For BF knowledge, the summary scores of nine items used at both baseline and process evaluation were compared. The overall mean score increased from 6.1 to 7.9 in AT-F between 2010 and 2012; there was a smaller increase in SF from 6.1 in 2010 to 6.9 in 2012 (results not shown). In relation to CF, improvement in knowledge on dietary diversity was larger in AT-F (from 61% to 93%) than in SF (from 63% to 86%) during the 2-year period (results not shown). During counselling observations, statistically significantly larger proportions of AT-F providers demonstrated positive interpersonal communication skills across all 13 indicators, compared with SF providers (Figure 2). Furthermore, AT-F providers exhibited an overall mean of 9.6 skills, whereas SF providers exhibited only 5.1 skills. More AT-F providers practised skills in negotiating behaviour change with mothers, such as encouraging mothers to identify a solution, obtaining commitment to try a recommendation and requesting mothers to repeat the advice received. Delivery of counselling sessions Table 5 presents the technical content covered during observed IYCF counselling sessions. AT-F providers delivered BF counselling services with statistically significantly more information and broader range of topics (7 out of 11 BF topics) than SF providers (three topics). During antenatal BF counselling, AT-F providers covered several individual topics more often than SF providers: the benefits of EBF for children (87.1% vs 45.5%) and mothers (64.5% vs 27.3%), the importance of good breast positioning and attachment (54.8% vs 0.0%), early initiation of BF (83.9% vs 27.3%), and the optimal BF practices (74.2% vs 18.2%). In sessions with mothers of infants under 6 months, only 16.7% of SF providers directly addressed clients concerns compared with 64.7% of AT-F providers. Counsellors in these sessions also checked more frequently whether mothers understood EBF (82.4% vs 33.3%), explained common reasons for breast milk insufficiency (41.2% vs 0%), and explained how to express and store breast milk (58.8% vs 0%). Although AT-F providers covered more information during the counselling sessions about CF than those in SF (5.5 vs 4.3 topics), the difference was only marginally significant (P < 0.10).

1014 HEALTH POLICY AND PLANNING Table 3 Structural attributes of A&T-supported facilities and standard facilities Indicators A&T-supported facility Standard facility Per cent/mean SD Per cent/mean SD Infrastructure and equipment a n ¼ 20 n ¼ 12 Had unshared counselling room (%) 65.0** 10.0 No. of equipment and supply items available (range 0 14) 8.7 1.0*** 3.9 1.9 Measuring equipment (%) Length board/height for children 80.0 50.0 Weighing scale for adults 100.0 70.0 Weighing scale for children 100.0 75.0 Kitchen utensils (for cooking demonstrations) (%) 100.0*** 35.0 No. of BCC materials (range 0 11) 5.1 0.9 4.8 1.9 Availability of services No. IYCF services offered (range 0 8) 7.2 0.7 7.0 1.7 % offering following services: Breastfeeding counselling in the third trimester of pregnancy 80.0 100 Delivery 100 91.7 Breastfeeding support during delivery 90.0 91.7 Breastfeeding counselling for children < 6 mo 100 91.7 Complementary feeding counselling for children 6 24 mo 100 91.7 Nutrition counselling for children 2 5 years 70.0 75.0 Food demonstration 95.0 91.7 Growth monitoring 80.0 66.7 Provider availability Staffing No. of medical doctors 1.0 0.6 0.9 0.3 No. of physician s assistants 2.1 1.1 2.3 1.3 No. of nurses/midwives 2.2 1.1 1.9 1.2 No. of village health workers/nutrition collaborators 9.4* 3.7 11.5 3.8 Total No. of staff 15.4 4.6 17.5 5.0 Time commitment of providers n ¼ 60 n ¼ 36 No. of days/month usually work at CHC 23.5 2.7 23.3 3.4 No. of hours/day usually work at CHC 8.3 1.2 8.5 1.5 In-service training b Staff received in-service training in IYCF in past 2 years (%) 86.7*** 13.9 No. of IYCF topics covered in training (range 0 17) 13.2 4.3*** 5.8 5.4 a Infrastructure items include tables, chairs, examination bed, medicine cupboard, document cabinet, clock, BCC materials holder, cooking demonstration module, food pyramid chart, hand washing basin, water container, plastic basin, refrigerator and white board. Other equipment and supplies include food/fruit/ vegetable demo, decorative items, playing mats and toy sets. IEC materials include individual and community growth chart, health cards/books, IYCF counselling card, leaflets, posters, educational pictures, breastfeeding facts/information, feeding infant guidelines, radio stories and media films. b A&T s in-service training package covers: behaviour change communication, individual counselling skills, group counselling skills, maternal nutrition, breastfeeding, breastfeeding positioning and attachment, expressing and storing breastmilk, common breastfeeding difficulties, complementary feeding, food demonstrations, feeding techniques, food hygiene and safety, building and assessing a menu, feeding sick children, breastfeeding for HIV-infected mothers, assessing nutrition status (growth monitoring), the Code Regulation on Marketing of Breastmilk Substitutes. *P < 0.05, **P<0.01, ***P<0.001. On average, AT-F individual counselling sessions were a few minutes longer than those in SF (22 26 vs 20 min), which corresponds to the results of more information covered during counselling sessions at AT-F. Mean duration for group counselling sessions was also longer in AT-F (41 43 minutes) than in SF (32 37 min) (result not shown), but these differences were not statistically significant. Outcome attributes Client satisfaction Client satisfaction with IYCF counselling was generally high, with no significant difference between facility types. At exit interviews, clients reported high levels of satisfaction with the counselling skills of health providers, usefulness of the advice received, facility appearance and BCC materials. Nearly all the

SOCIAL FRANCHISING IMPROVES NUTRITION COUNSELLING IN VIETNAM 1015 Table 4 Provider knowledge of IYCF practices in A&T-supported facilities and standard facilities Indicators A&T-supported facility (n ¼ 60) Standard facility (n ¼ 36) Per cent/mean SD Per cent/mean SD Knowledge of breastfeeding Provider knows: Babies should be breastfed immediately or within an hour of birth 98.3 100 Babies should be fed colostrum 98.3 100 Exclusive breastfeeding is only breastmilk < 6 mo 100 97.2 Mother should empty one breast before switching to the other 96.7 94.4 Even women with small breasts can produce enough breastmilk 96.7 91.7 It is possible to produce enough breastmilk when mother is not fed well 33.3* 11.1 Mother should continue breastfeeding if she is pregnant 76.7 74.3 Mother of a 4-month-old baby should not stop breastfeeding if the mother becomes ill 96.7 94.4 Mother should not give formula even when baby is not getting enough milk 96.7** 80.6 Baby < 6 mo should not be given water in hot weather 95.0*** 58.3 Mother should leave expressed breastmilk for babies ( < 6 mo) when she is away 96.7** 77.8 Total breastfeeding knowledge score (range 0 11) 9.9 1.0*** 8.8 1.2 No. of breastfeeding benefits to child that provider can name spontaneously (range: 0 6) 3.4 1.4* 2.8 0.9 No. of breastfeeding benefits to mother that provider can name spontaneously (range: 0 6) 4.4 1.8* 3.4 1.3 Knowledge of complementary feeding Provider knows: Complementary foods should be introduced at 6 8.9 mo 83.3 80.6 Minimum no. of times children aged 6 24 mo need complementary foods 90.0 80.6 Children aged 6 24 mo should eat 4 food groups 95.0 88.9 Can spontaneously name at least two iron-rich foods 86.7* 69.4 Total complementary feeding knowledge score (range 0 4) 3.6 0.7* 3.2 0.9 *P < 0.05, **P < 0.01, ***P < 0.001. clients reported that they would return to the same CHC for IYCF counselling in the future and would recommend the services to others. When asked open-ended about opinions on service quality during in-depth interviews, most of the franchise users praised the interpersonal skills and attitudes of the counsellors, describing them as enthusiastic, friendly and helpful. Another common theme among franchise users was the usefulness of the information received during counselling. One mother noted, I think [the franchise] is great and very helpful for pregnant women since it provides information that many never knew before. In my case, I was clueless when I had my first child, but the counsellor helped me during the counselling session with everything that I didn t know. This sentiment was shared by several first-time mothers who felt inexperienced in infant feeding. Similarly, the MTBT franchise was perceived as a good source for having information explained, raising concerns and asking questions. A mother explained, I find this information much more practical than other information I ve received. Sometimes I hear about it [child feeding] on television, but I can t understand it clearly. It s much more helpful to directly ask the counsellors what I want to know. Franchise users also highlighted the structural attributes of AT-F, praising the counselling room as clean, comfortable, bright, and appealing with decorations and posters. One mother described the comfort and privacy of an unshared counselling room, In such a room people can feel free to raise their questions. Normally they would hesitate to disturb others with their questions. Now, whenever we need to know anything, we go there for counselling. In general, very few negative opinions were shared; only three mothers commented that the facility was crowded or health staff were too busy. All of the franchise users reported that they would return to the facility, primarily to receive more information through counselling and ask questions as their children grow up. However, despite their willingness to return, a third of mothers said that they did not plan to do so in the near future. A few mothers believed that they needed to wait to receive another invitation card from a health worker in order to return to the franchise, revealing poor communication or misconception about service use. Several mothers perceived that there was no need to return unless their children was sick, whereas others reported being too busy or not wanting to go alone as reasons for not intending to return to the franchise.

1016 HEALTH POLICY AND PLANNING Figure 2 Provider interpersonal communication skills in A&T-supported facilities and standard facilities. Service utilization Figure 3 shows the percentage of the target populations in the communes who utilized AT-F services, according to the duration of franchise operation. As the process evaluation was conducted at the CHC level and very few mothers delivered their babies at CHC, data on EBF support counselling are not reported here. Although EBF promotion counselling reached about half of the pregnant women during their third trimester (48.9%), the overall utilization of AT-F services was low, reaching only 10% of the intended target population (results not shown). The rates of utilization dropped for service packages in later stages, down to only 1.4% among mothers with children 6 24 months of age. Utilization rates improved moderately with increasing duration of franchise operation, although the opposite was true for utilization of the CF education service, which was targeted to mothers with children at 4 6 months of age. Qualitative findings provide some insight into the barriers to franchise utilization. Among the 21 non-users interviewed in the A&T communes, 12 had never heard of the MTBT franchise, and two had heard of it but did not know anything else about the franchise. Other reasons for non-use included being too busy, the lack of perceived need to attend because the infant was well, insufficient knowledge about the services offered and crowded counselling room. Data on franchise service utilization are based on facility monitoring data and reports. Utilization data from SF are not included, given that SF do not have a data system for recording or reporting utilization of nutrition counselling. Discussion Our results provide evidence that incorporating elements of social franchising into government health facilities significantly enhances the quality of nutrition counselling services, particularly in the structure and process domains of service quality. Social franchising elements were built into the existing healthcare system without altering some fundamental aspects of service structure such as staffing composition and service and provider availability. However, health facilities supported by A&T were more likely to have designated counselling areas that were private, attractive and well-furnished with adequate equipment and materials. These improved structural attributes were the result of evident and uniform changes with high fidelity to the intervention design, particularly due to the specific inputs and procedural guidelines required for franchise setup. The largest differences between facility types were observed in the process attributes related to service provision, with

SOCIAL FRANCHISING IMPROVES NUTRITION COUNSELLING IN VIETNAM 1017 Table 5 Technical content provided during IYCF counselling sessions in A&T-supported facilities and standard facilities based on structured observation data Indicators A&T-supported facility (n ¼ 101) Standard facility (n ¼ 36) Per cent/mean SD Per cent/mean SD Counselling on breastfeeding during third trimester of pregnancy n ¼ 31 n ¼ 12 Explain the benefits of exclusive breastfeeding for the children 87.1* 45.5 Explain the benefits of exclusive breastfeeding for the mother 64.5* 27.3 Explain how breastfeeding works 51.6* 9.1 Explain the risks and hazards of not breastfeeding 41.9 18.2 Discuss the importance of skin-to-skin contact immediately after birth 25.8 0.0 Discuss the importance of good positioning and attachment 54.8*** 0.0 Use dolls or some things to illustrate the position of a baby at the breast 22.6 0.0 Talk about early initiation of breastfeeding 83.9* 45.5 Discuss the special properties of colostrum and why colostrum is important 83.9*** 27.3 Explain the optimal breastfeeding pattern 74.2** 18.2 Explain that no other food or drink is needed for the first 6 months only breastmilk 83.9 54.6 Total score (range 0 11) 6.7 2.6*** 2.6 2.1 Counselling on breastfeeding for mothers with children < 6 months n ¼ 34 n ¼ 12 Reinforce decision to breastfeed and how it will benefit both mother and baby 91.2 75.0 Check to be sure that the mother understands exclusive breastfeeding 82.4** 33.3 Discuss duration of breastfeeding; encourage exclusive breastfeeding 94.1 75.0 Monitor breastmilk intake 55.9* 16.7 Explain how breastfeeding works 52.9*** 0.0 Use dolls or some things to illustrate the position a baby at the breast 20.6 0.0 Explain to mother the steps of expressing and storing breast milk 58.8*** 0.0 Demonstrate to a mother how to cup-feed her baby safely 38.2 33.3 Explain the common reasons for insufficiency of milk 41.2** 0.0 Give appropriate advice on common breastfeeding difficulties 50.0 25.0 Address any concerns or problems the client might be having 64.7** 16.7 Total score (range 0 11) 6.5 2.4*** 2.8 1.3 Counselling on complementary feeding for mothers with children 6 24 months n ¼ 31 n ¼ 12 Explain the importance of breastfeeding while feeding solid/semi-solid foods 90.3 66.7 Encourage mother to breastfeed until 24 months or beyond 73.3 50.0 Explain different food groups that can fill in the gaps 100 91.7 Give advice for age-appropriate feeding including quantity, frequency 100 83.3 Explain safe preparation and storage of complementary food 83.3 58.3 Give appropriate advice on feeding skills (e.g. responsive feeding) 56.7 50.0 Give appropriate advice on feeding during and after illness 26.7 16.7 Give special advice for the children who are not receiving breastmilk 10.0 8.3 Conduct growth monitoring and referral of acute malnutrition 20.0 8.3 Total score (range 0 9) 5.5 1.8 4.3 2.1 *P < 0.05, **P < 0.01, ***P < 0.001. indications that AT-F provide higher quality counselling services than those of SF. More AT-F staff received training about IYCF and counselling skills than SF providers, and the training received was more comprehensive. Thus, their enhanced training likely contributed to AT-F providers demonstration of greater IYCF knowledge and interpersonal communication skills, which was corroborated by franchise users who reported on the useful information provided during counselling and positive interpersonal skills and attitudes of counsellors. Interestingly, both AT-F and SF providers knowledge improved compared with the baseline results, which may help to explain the small differences in some core technical knowledge about IYCF between the facility types. SF providers are trained on IYCF by NIN, A&T s partner franchisor, which may potentially account for their parallel improvements in knowledge over the past 2 years. Also, AT-F and the matched SF were located in the same districts, and there were some spillover effects of training content and materials. Still, there were large differences

1018 HEALTH POLICY AND PLANNING Figure 3 Percentage of target populations in the communes who utilized IYCF counselling services by duration of franchise operation. between AT-F and SF providers in relation to interpersonal communication skills demonstrated and the technical content delivered during counselling sessions, with AT-F providers exhibiting more skills overall. However, improved quality in the structure and process of AT-F services was not matched by higher utilization of franchise services. Utilization was satisfactory among pregnant women but very low for the other target populations, particularly mothers of children aged 6 24 months. Low utilization rates point to a need for improvement in generating demand for franchise services as well as addressing barriers to retain franchise service use. The reasons for not returning to the franchise and for non-use have important implications for demand creation strategies. For instance, one reason reported by both franchise users and non-users was the perception of not needing to attend because the child is well/not sick. Mothers should be motivated to attend the franchise even if their children are doing well, in order to receive advice and information that may be useful for keeping their children healthy and helping them to grow well. This perception among mothers may also help to explain the low utilization rates particularly for counselling services in later stages. When the mother does not perceive she has any problems or questions, she is unlikely to seek advice or counselling. This also reflects the general challenge in creating demand for preventive services, as compared with curative and recuperative services. Furthermore, utilization data for SF were not included in this phase of our research, limiting comparisons on this metric. The duration of franchise operation may be insufficient to affect service utilization, which also requires changing specific perceptions, beliefs and behaviours. Franchisees may require more time to overcome any pre-existing negative perceptions in the community about the quality of CHC services (Duong et al. 2005; Oanh et al. 2009) as the perception of quality is as crucial to service utilization as more objective measures of quality (Speizer and Bollen 2000; Bishai et al. 2008). Not all aspects of quality are easily perceived by users, who tend to focus on facility amenities, cleanliness and provider friendliness rather than the technical aspects of quality (Mendoza Aldana et al. 2001). To improve utilization, the quality improvements of services that carry the MTBT brand need to be communicated effectively to target populations. In addition, service fees are not yet incorporated into the MTBT franchise model, and this may affect users assessments of quality. Although fees may initially deter franchise use, introducing fees may serve to increase clients perceptions of the service value, as clients have a preference for affordable high-quality services over free services, which are associated with low quality (Ngo et al. 2009b). However, a challenge to fee collection has been the misconceptions of the Health Insurance Law and Government Decree No. 36, which regulate free health care (e.g. examinations and treatments) for children less than 6 years of age. Although nutrition counselling service is not included in these legislations, fee collection for any health service for young children is perceived by many as being in conflict with government policies. Despite overall low utilization rates, the high levels of client satisfaction with franchise services were corroborated by various data sources. However, there was a lack of statistical significance in the higher satisfaction among AT-F clients than that of SF clients, despite evidence of higher service quality in AT-F. When interpreting these results, we consider the potentiality of courtesy bias common in exit interviews. In Vietnam, clients tend to avoid expressing negative opinions about services. This would be the case especially while they are still at the facility, resulting in overly favourable responses (Glick 2009). Another consideration is heightened performance by providers when they know they are being evaluated, or the Hawthorne effect (Glick 2009). Clients may have experienced exceptionally positive counselling sessions on the days of the exit interviews. Furthermore, as with utilization, the short duration of franchise operation may partially contribute to the lack of difference in client satisfaction. Although improving the quality of care is an underlying goal of a social franchising model, our study tested this in the specific context of IYCF, which has not been addressed through

SOCIAL FRANCHISING IMPROVES NUTRITION COUNSELLING IN VIETNAM 1019 social franchising before, and in a context where service quality was relatively variable across facilities. We believe that the testing of this assumption was warranted, especially from an evaluation perspective, given that most AT-F fulfilled the minimum initial franchise requirements but there was still variability in initial conditions that might affect the success of the franchising approach. Compared with other systems-strengthening efforts, the model studied here brings in key social franchise elements of a brand identity, and explicit standards for service quality and for certification. However, as much as this critical element of social franchising is incorporated into the government health system, it is subject to the possible institutional and individual capacity limitations of the public sector itself. Still, the franchise-related elements of the model are those of explicit and uniform quality standards, backed up by mechanisms to assure those standards, thus offering the potential for ensuring programme quality and durability. An element of franchising that is not yet present in this programme model is the use of client costs to address costrecovery for the model. The current Vietnamese regulations for free health care for children <6 years of age place substantial limitations on the ability to levy user fees for the IYCF franchises. However, a separate costing study is currently underway, which will provide further information about start-up and implementation costs and provide insights into cost-related constraints to replicability and sustainability. Cost-recovery options that do not violate current Vietnamese regulations can also be assessed once the costing studies are completed. There are a few limitations to our study. The sample of health facilities was relatively small, limiting the power of statistical comparisons between facilities across various measures. For most domains assessed, however, statistical power was not an issue as many of the differences between AT-F and SF groups were statistically significant. The summary scores applied in this study are not standardized, as is the case of many scales, so it is difficult to interpret their relative differences and to know what magnitudes of difference really make an effect in terms of outcomes. For example, it is not known how much more knowledge of health workers is necessary to be skilled at influencing behaviour change or in translating their knowledge into improved maternal knowledge. Determining the meaningful size of differences in knowledge and other process-related quality indicators is an area of further research, particularly in different contexts. As discussed previously, another potential limitation is the duration of the intervention. Although the study was conducted 1 year after franchise implementation, not all of the franchises included in the study had been in operation for the full duration. Due to the small sample of facilities, we did not test the statistical significance of differences by duration of operation. However, it is possible that even a full 12 months of operations may be insufficient to have reached programme maturation (i.e. normalized service provision and uptake) and to assess some of the intervention outcomes. Our study has several strengths. First, it contributes to a small but growing literature on social franchising to improve healthcare quality in developing countries (Stephenson et al. 2004; Hennink and Clements 2005; Agha et al. 2007b; Bishai et al. 2008; Ngo et al. 2009b, 2010). To our knowledge, this is the first study to assess the impact of social franchising on the quality of nutrition consultation services. Second, the inclusion of intervention sites with matched controls provides stronger internal validity of our study results, in order to draw inferences about service quality changes that are indeed contributed by the MTBT franchise intervention. Third, the use of various data collection methods enabled an in-depth study of service quality across multiple dimensions. Finally, the study contributes to a larger process evaluation, which will help to inform how the different components and processes of the A&T intervention work to achieve impact. Although the impact on IYCF practices and child nutrition outcomes will be studied in a later impact evaluation, our study established that some intended elements of the implementation process are in place, reflecting fidelity to intervention design. Conclusions Health systems strengthening is increasingly viewed as a prerequisite to long-term, sustainable achievements in public health (Claeson et al. 2003). This is especially pertinent in Vietnam, which faces a decline in international donor funding for health programmes, as the country achieves middle income status (Mishra 2012). Building on the extensive coverage and vast network of health workers within the Vietnamese government health system, A&T s MTBT franchise intervention offers a feasible model that has demonstrated the potential to significantly improve the quality of IYCF counselling services at the primary care level, especially related to the structural and process attributes. However, the third aspect of service outcomes such as high rates of utilization also needs to be achieved. Improving all three attributes of quality of care is expected to translate into the programme goals of enhanced IYCF practices and better child nutrition. Competing interests The authors report no conflicts of interest. N.H. is the country programme director for A&T Vietnam and employed by FHI 360. N.H. was not involved directly in data collection or analysis, but provided inputs in the interpretation of findings. Supplementary data Supplementary data are available at Health Policy and Planning online. Funding This research was completed as part of the Alive & Thrive initiative in Vietnam. This program is managed by FHI 360 and funded through a grant from the Bill & Melinda Gates Foundation. References A&T. 2009. Feasibility report. Alive & Thrive, Ha Noi, Viet Nam.