How does power shape district health management team responsiveness to public feedback in low- and middle-income countries: an interpretive synthesis

Abstract Responsiveness is a core element of World Health Organization’s health system framework, considered important for ensuring inclusive and accountable health systems. System-wide responsiveness requires system-wide action, and district health management teams (DHMTs) play critical governance roles in many health systems. However, there is little evidence on how DHMTs enhance health system responsiveness. We conducted this interpretive literature review to understand how DHMTs receive and respond to public feedback and how power influences these processes. A better understanding of power dynamics could strengthen responsiveness and improve health system performance. Our interpretive synthesis drew on English language articles published between 2000 and 2021. Our search in PubMed, Google Scholar and Scopus combined terms related to responsiveness (feedback and accountability) and DHMTs (district health manager) yielding 703 articles. We retained 21 articles after screening. We applied Gaventa’s power cube and Long’s actor interface frameworks to synthesize insights about power. Our analysis identified complex power practices across a range of interfaces involving the public, health system and political actors. Power dynamics were rooted in social and organizational power relationships, personal characteristics (interests, attitudes and previous experiences) and world-views (values and beliefs). DHMTs’ exercise of ‘visible power’ sometimes supported responsiveness; however, they were undermined by the ‘invisible power’ of public sector bureaucracy that shaped generation of responses. Invisible power, manifesting in the subconscious influence of historical marginalization, patriarchal norms and poverty, hindered vulnerable groups from providing feedback. We also identified ‘hidden power’ as influencing what feedback DHMTs received and from whom. Our work highlights the influence of social norms, structures and discrimination on power distribution among actors interacting with, and within, the DHMT. Responsiveness can be strengthened by recognising and building on actors' life-worlds (lived experiences) while paying attention to the broader context in which these life-worlds are embedded.


Introduction
Responsiveness is one of the three health system goals, alongside health outcomes and fairness in financing introduced by the World Health Report of 2000 (WHO, 2000). Health system responsiveness has been judged necessary to provide inclusive, participatory and accountable services (Rottger et al., 2015;Askari et al., 2016). However, there is evidence that the public experiences difficulty in engaging with and eliciting responses from the health system (Golooba-Mutebi, 2005;Gurung et al., 2017). Furthermore, responsiveness is intended to draw attention to the needs of minority groups, but 'inequalities in responsiveness have received little attention' (Jones et al., 2011). While multiple public feedback mechanisms have the potential to enhance health system responsiveness (Molyneux et al., 2012;Cleary et al., 2017), there is limited information on their functioning and success in building system-level responsiveness-rather than on individual feedback pathways (Lodenstein et al., 2017;Whyle and Olivier, 2017).
This article presents an interpretive synthesis (Pope et al., 2007;Gilson, 2014) that addresses the overarching question: how do subnational health management teams receive, process and respond to public feedback? We sought first to identify whether and through what channels subnational health managers receive feedback from the public, how this feedback is analysed and whether responses to this feedback are generated and shared with the public. Second, because power has been cited as an influence on the responsiveness of health system agents (Berlan and Shiffman, 2012;Lodenstein et al., 2017), we included sub-questions related to power dynamics at the subnational level. We sought to understand how actors exercise power when receiving and responding to public • Applying actor interface analysis and Gaventa's power cube can help health policy analysts examine the interactions between structural influences and actor agency. This has value in understanding implementation challenges and in drawing out different dimensions of a goal as complex as health system responsiveness. • In a health system decision-making space such as the DHMT, power can be wielded in both positive and negative ways. How this power is exercised has a reinforcing effect on the public's sharing of feedback. Positive power practices support the generation of responses and even more feedback from the public. Negative power practices can limit generation of responses and the public's sharing of feedback. • Responsiveness could be strengthened by recognizing and building on the actor life-worlds that influence responsiveness practice. This could include leveraging politicians' power and personal interests while strengthening feedback channels to ensure meaningful public involvement and inclusivity and interventions to shape DHMTs' world-views and work environments to support responsiveness to public feedback. • Further research about power in the practice of health system responsiveness could test the conclusions and conceptual framework generated by this synthesis, in DHMTs in other contexts and in other spaces within the health system where decision-making on public feedback occurs.
feedback at the subnational level, why actors exercise power and what the effects of their power practices are. Research synthesis has a value in answering policy questions related to service delivery and organizational-and system-level change (Pope et al., 2007). As an interpretive synthesis, this article aims to draw out an understanding from the existing literature of whether and how power shapes responsiveness to public feedback at the subnational level and to consider what strategies might be deployed to deepen responsiveness. Synthesizing existing evidence also provides a platform for future empirical work to examine these issues more deeply (Gilson, 2014;Gilson et al., 2014a). Thus, the third aim of this work was to present conceptual insights drawn from the synthesized articles that could inform policy and research on health system responsiveness. The findings of this paper would be potentially relevant to policymakers, regional and district health managers, researchers and non-governmental organizations (NGOs) with an interest in promoting the inclusion of public input in shaping health systems. Our definition of health system responsiveness is how the health system reacts or responds to the public's needs and concerns (Whyle and Olivier, 2017). We understand the following processes as constituting the 'responsiveness pathway' within health system decision-making: receiving, processing (could include analysis, integration and/or prioritization) and responding to feedback (Whyle and Olivier, 2017). In this article, we focus on subnational health management teams 1 , which might be referred to as district health management teams (DHMTs) or Sub-county Health Management Teams (SCHMTs) depending on the country. We consider these teams to be a processing space in the health system where feedback could be received and acted on. 'Feedback' refers to the views, concerns and information shared by the public; 'feedback channel' or 'feedback mechanism' refers to how information, views and concerns from the public reach DHMTs. Feedback mechanisms might be formal or informal. Formal mechanisms are those that are legislated or provided for in policy and include 'community-level' feedback mechanisms such as health facility committees (HFCs), intersectoral health forums or community monitoring (Molyneux et al., 2012) and 'individual-level' feedback mechanisms such as suggestion boxes (Atela, 2013), exit surveys and incident reports (Khan et al., 2021). Informal mechanisms are not necessarily mandated or legislated and might appear in contexts where formal mechanisms are absent or are considered ineffective by citizens (Tsai, 2007;Hossain, 2009;Lodenstein et al., 2018). Informal mechanisms include individual complaints or compliments shared directly with frontline providers and health managers or via an intermediary and collective feedback such as public protests or 'public buzz' (conversations in public places) (Hossain, 2009;Lodenstein et al., 2018).
There is increasing attention to the complex roles DHMTs play in managing and leading health systems at the district level in low-and middle-income countries (LMICs) (Kwamie et al., 2015;Nyikuri et al., 2017;Bulthuis et al., 2021). However, there is little evidence about how public feedback is brought into DHMTs' decision-making and of the influences on these processes. Although power is at the 'heart of every policy process' (p 361) (Erasmus and , including health system responsiveness (Berlan and Shiffman, 2012;Lodenstein et al., 2017), there are few purposeful examinations of power in health policy and systems research (HPSR) (Gilson and Raphaely, 2008;Topp et al., 2021) and even fewer examinations of power in the practice of health system responsiveness (Khan et al., 2021). To strengthen responsiveness at the subnational level, a better understanding of how public feedback is handled within decision-making spaces such as the DHMT (including the influence of power) is important.
In the 'Methods' section, we describe the approach we adopted in conducting this work. We then present our findings in two parts: the first is a description of the various ways in which public feedback is received and responded to by DHMTs, and the second is a synthesis of the power dynamics influencing how DHMTs receive and generate responses to public feedback.

Methods
We conducted a purposive review and interpretive synthesis (Thomas and Harden, 2008;Gilson, 2014). Interpretive synthesis allows researchers to draw conclusions on the collective meaning of pooled studies in a systematic manner (Gilson, 2014). This approach, also used more widely in HPSR (Erasmus, 2014;Gilson et al., 2014b;Parashar et al., 2020a), draws on studies that did not consider the review question and generates new interpretations of reported study experiences by going beyond the original studies during analysis (Pope et al., 2007;Thomas and Harden, 2008). We have drawn on the enhancing transparency in reporting the synthesis of qualitative research guidelines (Tong et al., 2012) and in reporting our synthesis methodology (see Supplementary Material 1).

Data sources search strategy and screening
The search for papers was conducted on PubMed, Google Scholar and Scopus between December 2020 and March 2021 using the search criteria as presented in Table 1. The databases were chosen because they were free access and comprehensive and are known to cover health-related matters. A total of 694 papers were identified through database searches. NK made all the searches in consultation with a librarian. All the citations from the different databases were exported to District health management team*, sub-county health management team*, district health manager*, regional health management team*, regional health manager*, provincial health management team, provincial health manager* a The two groups were ultimately combined with AND.
Excel, and duplicates were removed. This was followed by screening of the title and abstracts for relevant papers (Supplementary Material 2). The eligibility of the studies selected was discussed with three members of the authorship team. Hand searching of the reference lists of articles identified was used to identify additional articles judged relevant to the review and synthesis questions. In total, 703 papers were identified.

Eligibility criteria and quality appraisal
Articles were included in this review if they met the following criteria: (1) they contained substantial content on DHMTs receiving, processing and/or responding to public input; (2) they focused on LMICs; (3) they were in English and (4) they were published between 2000 and 2021. The latter criterion was adopted because responsiveness was introduced as a health system goal by the World Health Organization (WHO) in 2000 (WHO, 2000). Twenty-one articles were retained after screening. Figure 1 summarizes the screening process. Selection of the articles included in the review combined assessment of specific relevance (empirical analyses of district health managers' experiences with public feedback, views and concerns) with quality. We adopted the checklist in Supplementary Material 3, drawn from the study by Dixon-Woods et al. (2006) to assess the quality of the included studies. None of the 21 studies was excluded following the quality appraisal.

Synthesis methodology
We adopted a framework approach to analysis (Walt and Gilson, 2014;Parashar et al., 2020a) drawing on Gaventa's power cube (Gaventa, 2005) and Norman Long's concept of actor life-worlds (Long, 2003). Our power analysis was informed by the understanding that power is a dynamic resource that can be shared and used by individuals and groups (Veneklasen and Miller, 2002;Gaventa, 2005). Gaventa's power cube was a good fit, given its relevance for researchers with applied interests and as we hoped through our power analyses to generate ideas about how responsiveness might be deepened. Norman Long's actor-oriented perspective on power illustrates how the lived experiences of actors, their interactions and power struggle shape policy implementation (Long, 2003). The combination of these two power frameworks supported analysis that both (1) identified structural and organizational power (Gaventa's power cube) and (2) considered power at the micro level to understand power differentials and struggles between actors (Long's actor interface analysis) and how both impacted the practice of responsiveness. We focused on actors in a bid to be responsive to calls for more actor-centric HPSR (Sheikh et al., 2014;Topp et al., 2021). Table 2 presents a summary of Gaventa's power cube and illustrates the three dynamic and interacting dimensions of power: levels, spaces and forms of power (Gaventa, 2003(Gaventa, , 2005. Spaces for power refer to mechanisms or channels where actors can influence decisions or policy. These spaces are shaped by power relations that determine who can participate in them (Gaventa, 2003(Gaventa, , 2005. Levels of power include local, national and international arenas. The forms of power build on Lukes' 1974 three dimensions of power (Lukes, 1974(Lukes, , 2004 and encompass visible, hidden and invisible forms of power. Long's actor interface analysis supported an in-depth exploration of power struggles between actors (Long, 2003;Parashar et al., 2020a). According to Long, the points of interaction between actors in relation to a policy can be understood as actor interfaces. These interfaces are shaped by intersecting 'actor life-worlds', a term that refers to the lived experiences of actors. The formation of these life-worlds is dynamic and linked to the contexts of actors' lives (Long, Long, 2003). Table 3 presents a summary of these contexts including their associated elements. The contexts include knowledge and power relationships in society and organizations, personal characteristics and world-views influenced by social-cultural-ideological standpoints.
Power practices ranging from domination, collaboration, negotiation and resistance to contestation may be observed within the actor interfaces (Long, 2003;Parashar et al., 2020a). Table 4 elaborates more on these power practices. Concerning Gaventa's power cube, we anticipated that these power practices may be observable across the forms and within the spaces and levels of power.

Data extraction and derivation of themes
We first read and re-read the studies to identify raw data for the synthesis. A data extraction Excel sheet was devised to assist in systematically identifying characteristics of the papers, study objectives and actors described in the papers. The template for extraction of content from the review articles also included columns for the feedback channel, the Decision-making at the subnational level might include counties, districts and provinces down to the community level Source: Gaventa, 2005;Gaventa and McGee, 2013. content of feedback, processing of feedback, responses generated from feedback and composition of the DHMT (see Supplementary Material 4 for the full list of articles and sample of extracted content). This content was useful to answer the overarching review question. Texts were also uploaded onto Nvivo version 12, to support line-by-line coding of the primary texts. For the power synthesis, we drew on concepts from Gaventa's power cube and Long's interface analysis to code for data on actors with whom the DHMT interacted in receiving and responding to feedback, spaces and levels where feedback was received, discussed and responded to, forms of power observed within the DHMT or influencing the DHMT in receiving and responding to feedback, power practices by individuals or groups of actors, effects of power practices and actor life-worlds underpinning practices of power. Data for actor life-worlds were obtained by coding for actor life-world dimensions and then sub-coding for the characteristic elements of actor life-worlds described in Table 3.  Actors work together to support an action or decision Contestation Opposition between two actors interacting at an interface Resistance Actors object to or oppose a decision or action of another actor Source: Long, 1999;Parashar et al., 2020a. During coding, we considered data from all sections of an article, including author judgements (author's insights into reported data) (Gilson, 2014). This included information on context reported by the authors that was useful for our understanding of findings on power. The process of deriving themes combined deductive and inductive approaches. Themes were developed drawing on the conceptual and power framework, and all studies were coded according to which element of the frameworks they addressed. In addition, new topics were developed and incorporated as they emerged from the reviewed articles. To support comparison across papers, data extracted from various sections of the primary studies were entered into charts. NK developed a written summary to accompany the charts for discussion and agreement with the authorship team. Analysis of the evidence presented in the charts formed the basis for an overarching synthesis about how power might influence the functioning of a space within the health system where public feedback is received and responded to.

Study scope
This synthesis was limited to papers that discussed receiving and/or responding to public feedback by the DHMT. It is constrained by the limits of the included papers. For example some of the revwied articles did not alwsy link public feedback to a response generated at DHMT level. The paper focuses on the practice of responsiveness by DHMTs and power dynamics between multiple actors influencing DHMT responsiveness; other factors such as the design of responsiveness policy and guidelines are not included here, although we recognize that these may influence DHMT handling of public feedback.

Characteristics of the articles
The 21 articles reported studies that mainly used qualitative data collection methods such as in-depth interviews, focus group discussions, observation and document review. The studies formed two broad categories: those that examined health system functioning with some consideration of public feedback at the district level (Kapiriri et al., 2003;Tuba et al., 2010;Maluka, 2011;O'Meara et al., 2011;Cleary et al., 2014;Van Belle and Mayhew, 2016;Nyikuri et al., 2017;Tsofa et al., 2017;McCollum et al., 2018;Henriksson et al., 2019;Razavi et al., 2019;Jacobs and Baez Camargo, 2020;Mukinda et al., 2020;Parashar et al., 2020b) and intervention studies that reported on efforts to enhance inclusion of and response to public feedback in the priority setting Byskov et al., 2014;Zulu et al., 2014), including through social accountability approaches (Blake et al., 2016;George et al., 2018;Boydell et al., 2020;Butler et al., 2020). The reviewed studies reported on experiences from a range of geographical contexts spanning sub-Saharan Africa (18 of 21 papers), India (2/21) and Central Asia (Tajikistan) (1/21) and addressed a range of issues from general health governance to specific service delivery areas (see Figure 2).
Regarding the governance contexts in which the DHMTs are operated, 15 out of 21 articles mentioned a decentralized context. However, in the majority of these 15 studies, there was inadequate detail to judge the form of decentralization, with only six studies, three in Kenya (Nyikuri et al., 2017;Tsofa et al., 2017;McCollum et al., 2018) and three in Uganda (Kapiriri et al., 2003;Razavi et al., 2019;Boydell et al., 2020), clearly stating and providing details of a devolved context.

Results
The results of the literature review and synthesis are presented in two broad parts. The first part describes the processes of receiving, processing and responding to public feedback at the DHMT level, including specific feedback channels utilized by the public and the content of public feedback. The second part focusses on the exercise of power by the DHMTs themselves and actors with whom DHMTs interacted. Concerning how DHMTs managed public feedback, findings from the review suggest that a mix of formal and informal channels was utilized to receive public feedback, but there was little analysis (or processing) of feedback. Feedback channels in the reviewed studies appeared to exclude vulnerable groups, and in the few instances where responses were generated, there was little communication to the public. These elements of responsiveness are presented in more detail in subsequent sections.
Processes of receiving, processing and responding to public feedback by DHMT How DHMTs received feedback from the public? From the studies, we identified five broad categories of channels through which DHMTs received feedback from the public (Box 1). Four of these categories were formal mechanisms established in country policy and guidelines. The last category, informal feedback channels, was more commonly reported in contexts where challenges were faced in the functioning of the formal mechanisms.
Despite policy provisions, several studies reported variations in the extent to which public feedback successfully reached DHMTs (Kapiriri et al., 2003;Maluka, 2011;Nyikuri et al., 2017;McCollum et al., 2018;Razavi et al., 2019). Poor attendance at budgeting and planning meetings by community members was cited as a challenge to including public feedback in the priority setting (Kapiriri et al., 2003;Maluka, 2011;McCollum et al., 2018;Razavi et al., 2019). In Kenya, a lack of capacity and clarity about who was responsible for budgeting and planning within the department of health in the newly decentralized context constrained inclusion of public priorities (Nyikuri et al., 2017). In Ghana, the absence of 'functioning' mechanisms within the district bureaucracy combined with a focus on vertical (to regional managers) and horizontal (to NGOs) accountability limited public accountability (Van Belle and Mayhew, 2016). Similarly, in South Africa, there was a predominance of internal bureaucratic accountability initiatives focused on the performance of health-care providers at the expense of accountability to the public (Mukinda et al., 2020). Finally, in Tajikistan, NGO-supported community-based organizations (CBOs) at the village level had little leverage to demand feedback from the DHMT as they were directly linked to NGOs rather than the state mechanisms (Jacobs and Baez Camargo, 2020).

Who provided feedback and what was the content of the feedback?
The equity element of responsiveness requires consideration of which groups provide feedback and whether marginalized groups give feedback (WHO, 2000;Khan et al., 2021). However, in the majority of papers reviewed, feedback was commonly reported as though voiced by a homogenous public. Several studies noted that vulnerable groups were often left out of priority-setting processes for the health sector (Kapiriri et al., 2003;McCollum et al., 2018;Razavi et al., 2019), lacked representation in decision-making committees (Van Belle and Mayhew, 2016) or experienced barriers to voicing concerns about specific services such as reproductive health (RH) (Boydell et al., 2020). These vulnerable groups were women, the youth, people with disability and adolescents (Kapiriri et al., 2003;Van Belle and Mayhew, 2016;McCollum et al., 2018;Boydell et al., 2020). Four studies explored in some detail the factors that contributed to the exclusion of vulnerable groups in terms of priority setting (Kapiriri et al., 2003;Van Belle and Mayhew, 2016;McCollum et al., 2018;Boydell et al., 2020). This is discussed in more detail in the section on power.
Of the 21 articles reviewed, only six included details about the content of public feedback (see Supplementary Material 5). Drawing on these papers, we identified four broad categories of public feedback: i) provider-client interactions, ii) infrastructure, staffing and commodity-related issues, iii) requests for the introduction of new services and iv) broader environmental and health system issues impacting health service uptakes.

Processing of public feedback
About a third (7/21) of the studies reported some form of analysis or consolidation of feedback at the district level O'Meara et al., 2011;Byskov et al., 2014;Zulu et al., 2014;Blake et al., 2016;George et al., 2018;Butler et al., 2020). The details of processing public feedback in the reviewed studies are summarized in Table 5, which highlights that in the cluster of health sector priority-setting studies (Maluka, 2011;Maluka et al., 2011;O'Meara et al., 2011;Byskov et al., 2014;Zulu et al., 2014), we identified consolidation of community input at the facility level and then upward submission to the district level. Table 5 also highlights that practice differed from recommendations about processing arrangements. For example, review by a multi-stakeholder board was uncommon in Tanzania (Maluka, 2011), but public appeal of disseminated priorities hardly occurred across several countries (  Social accountability interventions supported by NGOs (1) Community scorecards (George et al., 2018) (2) Facility report cards (Blake et al., 2016) (3) Community dialogue meetings (Butler et al., 2020) (4) CBOs/village organizations (Jacobs and Baez Camargo, 2020) Informal feedback mechanisms (1) Direct calls to DHMT members (Van Belle and Mayhew, 2016) (2) Phone calls to influential actors (Parashar et al., 2020b) (3) Public airing of service delivery concerns on radio (Van Belle and Mayhew, 2016) Finally, Table 5 shows the processing of public feedback in the cluster of studies reporting on social accountability interventions. This processing was supported by NGOs and mainly entailed 'quantitative analyses' of facility scorecard results (Blake et al., 2016) and village-level report cards (George et al., 2018) to develop summaries of data collected from service users. In two studies, conducted in Malawi (Butler et al., 2020) and Uganda (Boydell et al., 2020), feedback from multiple mechanisms was integrated, combining both qualitative and quantitative analyses. Across all four studies describing social accountability interventions, public feedback was shared with district health managers (Blake et al., 2016;George et al., 2018;Boydell et al., 2020;Butler et al., 2020), who responded as described later. Notably, processing of feedback was not done within the DHMTs in these studies. Instead, NGOs performed the analysis (Table 5) and shared the findings with the DHMT.

Responses to public feedback
Seven studies discussed some detail on responses to public feedback. One study conducted in Zambia highlighted district managers' 'selection' of issues to respond to, based on their perception of what they could influence. For example, there were instances when DHMTs simply 'took no action' despite receiving public feedback. This was reported in the study by Tuba et al. (2010) regarding complaints related to waiting times and health provider behaviour such as rudeness to the public (Tuba et al., 2010). However, the same district managers responded to complaints about overpriced nets at the facility level by collaborating with an NGO to set up a monitoring system for tracking the sale of insecticide-treated nets (Tuba et al., 2010). In Ghana, the DHMT also 'took no action' in response to public feedback despite the public's efforts to express their service delivery concerns through radio and calls to DHMT members (Van Belle and Mayhew, 2016). Across both studies, there was a failure to acknowledge complaints from the public, and thus, no responses were generated at all. In the study by O'Meara et al., response to public feedback was in the form of community priorities being adopted only if they aligned with national targets (O'Meara et al., 2011). All other priority-setting studies (Maluka et al., 2010;2011;Byskov et al., 2014;Zulu et al., 2014) simply did not discuss whether community priorities eventually informed district plans.
Four other studies highlighted specific responses generated at facility, community or district levels. In these studies, the reported responses appeared to have had system-level effects. They included provision of a vehicle to improve the referral system within the district (Blake et al., 2016), increasing budget allocations for family planning (FP) and RH services (Boydell et al., 2020), inclusion of identified service needs in the financial plan for the subsequent year (George et al., 2018) and improvements in facility infrastructure and initiation of service delivery in defunct facilities (George et al., 2018). These four studies also reported escalating some feedback to the regional and national level, but responses from these higher levels were not discussed.

Manifestations of power in processes of receiving and responding to public feedback at the district level
In this section, our findings related to the exercise of power are presented in three subsections. First, we consider where actor interfaces (points of interactions between actors) are situated and how power was exercised within and across Gaventa's levels and spaces of power. Second, we explore the forms of power observed at the actor interfaces, including their linked power practices, and the actor life-worlds underpinning the identified forms and practices of power. This approach allowed further deconstruction of the exercise of power to reveal the agency and motivations of actors. Third, we present findings on the effects of the observed power dynamics on DHMT handling of public feedback. All these findings are summarized in Table 6, which presents synthesis findings about the processes of receiving and responding to public feedback. For the various instances drawn from the reviewed articles in Table 6, we highlight the observed form of power, the level and the space where this power was observed to be exercised. Table 6 also presents the associated practices of power, underpinning actor life-worlds and the effects on responsiveness for each of the instances highlighted.

Studies
Details of proposed processing of feedback received from the public

How processing played out in practice as reported in reviewed articles
Priority-setting studies Maluka, 2011;Maluka et al., 2011;O'Meara et al., 2011;Byskov et al., 2014;Zulu et al., 2014 Consolidation of community priorities shared from the community level, upwards to facility and district levels Community priorities were consolidated and shared upwards to the PHC facility level and then to the district level Maluka, 2011 'Review by a multi-stakeholder board' comprising community representatives to check for inclusion of community priorities This board was often bypassed because they did not meet frequently. (1) The board also lacked the capacity to scrutinize budgets and plans for the inclusion of community priorities O' Meara et al., 2011 'Community priorities were considered in relation to district targets' (which were shared in a top-down process informed by national indicators) The community priorities were excluded if they did not align with the national indicators and district targets. District targets were developed in a separate process that was linked to national indicators Maluka et al., 2011;Byskov et al., 2014;Zulu et al., 2014 'Information provision to the public to give room for appeal' before formally adopting the district plans The public did not appeal any of the proposed priorities shared Social accountability studies Blake et al., 2016;George et al., 2018 Quantitative analyses of facility and community scorecards results Blake et al., 2016;Butler et al., 2020 Combination of quantitative and qualitative summaries of findings from multiple feedback mechanisms Table 6 highlights the multiple interactions between DHMTs and various actors in the processes of receiving and responding to public feedback. These actors included: community representatives, individual community members, political actors, regional and national health managers and NGOs. At these points of interaction, we identified actor interfaces situated both within and across Gaventa's levels and spaces of power. Importantly, despite having a formal mandate to oversee health service delivery and planning, many DHMTs, even in decentralized countries, had limited decision-making autonomy. At the interface between DHMTs and regional/national health managers, the higher-level managers often 'dominated' the planning process. For example, DHMTs could not make final decisions on plans and budgets at the district level as they were required to follow national-level guidelines, with little room for local priorities. Changes to district plans were also often made at the regional or national level (Maluka, 2011;O'Meara et al., 2011;McCollum et al., 2018;Henriksson et al., 2019). DHMTs also operated in contexts of resource scarcity illustrated by unpredictable and inadequate disbursements of funds from national or regional levels (Van Belle and Mayhew, 2016;Nyikuri et al., 2017;Jacobs and Baez Camargo, 2020), understaffing and low supplies of commodities for the primary health-care (PHC) facilities they supervised (Tuba et al., 2010;Jacobs and Baez Camargo, 2020). NGOs operating at the district level sometimes filled a few of these resource gaps (Tuba et al., 2010;Van Belle and Mayhew, 2016;Jacobs and Baez Camargo, 2020), forming an interface with the DHMT. However, there were drawbacks related to NGO 'collaboration' with DHMTs. For example, in Ghana, Van Belle and Mayhew reported that the DHMT in the study districts engaged with three NGOs frequently, leaving little opportunity for engagement and inclusion of the public in planning activities (Van Belle and Mayhew, 2016).

Multiple actors and a wide range of interactions across health system levels and spaces in relation to receiving and responding to public feedback
In several studies, we identified clusters of DHMT members working closely together (Maluka, 2011;Zulu et al., 2014;Van Belle and Mayhew, 2016;Jacobs and Baez Camargo, 2020), which we judged to be 'closed spaces'. Here, decisionmaking occurred with little or no consultation with other DHMT members and stakeholders. These 'core teams' comprised individuals with leadership roles in the DHMT or with resource allocation-related roles. In two priority-setting studies, these core teams dominated health planning by withholding access to district plans such that there was inadequate time for other DHMT members' or stakeholders' views to be incorporated into the plans before upward submission to the national level (Maluka, 2011;Zulu et al., 2014). In one of the broader governance studies in Tajikistan, a core team 4 within the DHMT concentrated resources at the district hospital and denied other DHMT members' resources for their activities, including for visiting peripheral facilities where they could have picked up issues related to public feedback (Jacobs and Baez Camargo, 2020). At many of the interfaces shown in Table 6, the public was often the less powerful actor. However, they were not passive actors; when the 'invited spaces' failed to provide an avenue for public feedback to reach the DHMT, the public attempted to evolve 'claimed spaces' where they voiced complaints. For example, in Ghana, where DHMTs were more focused on reporting upwards to their regional managers and horizontally to NGOs, the public used radio and increased litigation to share complaints and concerns about the health system (Van Belle and Mayhew, 2016). In Tajikistan where autocratic rule had undermined formal voice mechanisms at the interface between health providers and the public, the public provided in-kind contributions at under-resourced peripheral facilities, creating a degree of answerability for service provision between the community and frontline providers (Jacobs and Baez Camargo, 2020). These claimed spaces seemed to have mixed results in tilting power towards the public. In the Ghanaian study, the authors reported that despite the public's efforts to share feedback in new ways, such as radio, and through direct calls to DHMT members, there was a failure to acknowledge this public feedback by the DHMTs who failed to respond (Van Belle and Mayhew, 2016), while in 536 Health Policy and Planning, 2023, Vol. 38, No. 4        the Tajikistan study, the authors observed that it was possible that the public may have been coerced by frontline healthcare workers (HCWs) into providing in-kind contributions that reportedly contributed to higher facility-level responsiveness to the concerns of the public (Jacobs and Baez Camargo, 2020). In the study by Parashar et al., the public was more successful with their claimed space, as they leveraged connections to powerful and influential actors ('power relationships') to access their entitlements as beneficiaries in a mother-child safety programme (Parashar et al., 2020b).

Forms of power and their linked power practices were underpinned by varying and interacting actor life-worlds in relation to receiving and responding to public feedback
In this section, we explore the various forms of power and specific practices of power identified from the reviewed studies. Furthermore, we also present what actor life-world supported the identified exercise of power. These instances of exercise of power are highlighted in Table 6, which shows that visible power was a dominant form of power associated with both positive and negative power practices. For example, the DHMT, given its formal mandate and managerial authority over health planning and service delivery, was a space where 'visible power' was commonly exercised. Some of the power practices linked to visible power include, for example, in the intervention studies implementing social accountability initiatives, 'collaboration' between NGOs and DHMTs, as both exercised visible power to identify and respond to public feedback. NGOs drew on their resources and technical expertise to support the functioning of community scorecards (Blake et al., 2016), facility report cards (George et al., 2018), village health teams (VHTs) and local civil society organizations (CSOs) (Boydell et al., 2020;Butler et al., 2020) to collect public feedback, while DHMTs exercised their positional power to respond to some of the issues raised by the public (see the section on 'Responses to public feedback'). However, there were also instances where DHMT members used their power in ways that undermined responsiveness to public feedback. For example, in Ghana, DHMTs 'dominated' the public and community representatives by failing to acknowledge public feedback despite the public's efforts to use new channels like radio, litigation and direct calls to the DHMTs to share their concerns (Van Belle and Mayhew, 2016).
In several studies, 'visible power' flowed in a top-down manner and the DHMT was commonly 'dominated' by national O'Meara et al., 2011;Zulu et al., 2014), regional  and political actors (Nyikuri et al., 2017;Tsofa et al., 2017;McCollum et al., 2018;Razavi et al., 2019). At the national/DHMT, regional/DHMT and politicians/DHMT interfaces, we noted domination underpinned by 'relationships of power' rooted in organizational hierarchy (Maluka, 2011;O'Meara et al., 2011;Byskov et al., 2014;Zulu et al., 2014) and control over resources (Maluka, 2011;Nyikuri et al., 2017;Tsofa et al., 2017;Razavi et al., 2019). For example, in Kenya, there was reportedly little inclusion of sub-county health managers and the public in the health priority setting despite recent decentralization (Nyikuri et al., 2017;Tsofa et al., 2017;McCollum et al., 2018). Decentralization had created semi-autonomous counties headed by political leaders. At the interface between health managers and politicians, both county and sub-county health managers had little room to challenge decisions made by politicians or their appointees (McCollum et al., 2018). SCHMTs (DHMT equivalent) also experienced significant resource constraints, which made it difficult for them to learn about public feedback at the facility level (as they could not conduct timely support supervision visits) or in stakeholder meetings (Nyikuri et al., 2017;Tsofa et al., 2017;McCollum et al., 2018). In this case, SCHMTs (and the department of health) were 'dominated' by higher-level county actors who concentrated resources at the county level.
We also identified 'hidden power' sometimes influencing whose (and what) feedback DHMTs received. In the prioritysetting studies, powerful actors controlled public participation processes. In Uganda, at the public/politicians interface, politicians exercised hidden power by selectively mobilizing rich community members, while the youth and poorer community members were invited only after decisions regarding project costs and plans had been made (Kapiriri et al., 2003). In Kenya, despite having the mandate to mobilize all community members for public participation, politicians made little effort to educate the public on their rights to participate in the priority setting and how to do so, perpetuating low public awareness and participation in the priority setting (McCollum et al., 2018). In these exercises of hidden power, politicians commonly 'dominated' the public, a power practice underpinned by two interacting life-worlds. One was 'power relationships' rooted in politicians' positions of authority and access to information. Second was the 'personal concerns' of politicians who wanted to appeal to their voter base and retain political power. In a Kenyan study, politicians reportedly prioritized resource allocation to areas where they had political support to secure votes or repay political promises (McCollum et al., 2018). Similarly, in a social accountability intervention study reporting findings from Uganda, local politicians were perceived to sweep in to claim credit for changes arising from public feedback to garner political recognition (Boydell et al., 2020).
In two priority-setting studies, the public reacted to domination by politicians 'with resistance and contestation'. In Kenya and Uganda, the public perceived that their participation was tokenistic and resisted attendance of public participation meetings scheduled by local politicians (Kapiriri et al., 2003;McCollum et al., 2018). We judged this resistance to be underpinned by life-worlds shaped by 'ideological worldviews and personal characteristics'. For example, in Kenya, one of the marginalized communities held the belief (worldview) that public participation would not change the community's circumstances given the 'historic neglect' of their region 5 (McCollum et al., 2018). In Uganda, beliefs of exploitation by politicians were linked to a view that politicians were paid to conduct public participation meetings, but they (politicians) then failed to pay the attendees. This contributed to low attendance of the public participation meetings by youth, the majority of whom was unemployed and felt that there should be tangible benefits from public participation (Kapiriri et al., 2003). In addition, there were reportedly greater efforts by local politicians towards public mobilization during election periods, compared with the poor mobilization done for health sector planning. This resulted in feelings (personal concerns) among the public of 'being forgotten' by politicians after elections, which in turn underpinned their resistance to participation in public meetings for the priority setting.
'Invisible power' also appeared to influence both receiving and responding to public feedback. Constraints to receiving feedback that demonstrated invisible power included structural issues such as people's illiteracy, lack of interest and awareness about the possibilities of participation Zulu et al., 2014;McCollum et al., 2018), poverty and unemployment (Kapiriri et al., 2003) and a culture of not questioning those in authority . In three studies, we also identified the influence of patriarchal norms in keeping women and youth from providing feedback on priority-setting (Kapiriri et al., 2003;McCollum et al., 2018) and RH services (Boydell et al., 2020). In Uganda, men within the community perceived that women's and youth's participation in decision-making processes was 'rebellious' even though policy guidelines specifically identified women and youth as vulnerable groups whose views were to be included in all policy processes (Kapiriri et al., 2003). Women failed to attend local council meetings because they could not afford to dress 'appropriately' and look 'presentable' at these meetings (Kapiriri et al., 2003). In Kenya, women were often busy with household chores when public participation meetings were planned, and even when they attended, lacked the confidence to speak (McCollum et al., 2018). In the Ugandan study by Boydell et al., women and youth agency in accessing and providing feedback about FP services was compromised by patriarchal and moral views that opposed FP use (Boydell et al., 2020). In a different context, in Tajikistan, a history of autocratic leadership (characterized by absent electoral processes and local-level formal voice mechanisms) contributed to low expectations of answerability from local officials and district health managers (Jacobs and Baez Camargo, 2020). As a result, there was no attempt by the public to provide feedback to the DHMTs at all.
Concerning responding to public feedback, we identified the invisible power of bureaucratic hierarchy illustrated first, by a culture within DHMTs of adopting top-down priorities (Kapiriri et al., 2003;Maluka et al., 2011;O'Meara et al., 2011;Byskov et al., 2014;Zulu et al., 2014). This culture persisted despite decentralization to the district in all of the priority-setting studies' contexts. Kapiriri et al. (2003) described this as a tendency to plan 'for' the community rather than 'with' the community retained from the previously centralized health system (Kapiriri et al., 2003). Second, two studies highlighted the focus of health managers and providers on internal performance requirements and horizontal accountability relationships (with NGOs) at the expense of responses to public feedback (Van Belle and Mayhew, 2016;Mukinda et al., 2020). In South Africa, Mukinda et al. (2020) identified 19 formal and informal accountability mechanisms targeting district-level health managers and providers, the majority of which was related to performance accountability 6 (Mukinda et al., 2020).
From examining life-worlds (Table 6), it appears that ideological world-views mirror the exercise of invisible power by shaping actors' views of what is acceptable. In several studies, the power practices DHMTs demonstrated in receiving and responding to public feedback were underpinned by beliefs and values, an element of ideological world-view. For instance, in Zambia, DHMT members failed to respond to community concerns related to discrimination in waiting times, due to a belief that 'it was fair that waiting times differed for different types of people' (p 6) (Tuba et al., 2010). One manager noted: Well in society, we have different people. Like even politicians can't go in the queue. So that's how you find, when people see that, they will start complaining. But it's because maybe of one's status in society, for example, the xxx (referring to a political position in the district) and other political leaders (p 6) (Tuba et al., 2010). In the same study, decision-makers did not recognize the public as legitimate stakeholders because they lacked technical training (Tuba et al., 2010). This together with the managers' views previously suggests little value for public feedback. In contrast, in another Zambian study (Zulu et al., 2014), DHMT members held values of openness and transparency that promoted collaboration between the DHMT and action research team to improve inclusivity in the priority setting. The DHMT drawing on their motto of 'provision of health services in partnership with the community' was able to quickly revive feedback channels such as HFCs, which had not been functioning before the intervention to enhance the inclusive priority setting (Byskov et al., 2014;Zulu et al., 2014). This DHMT's world-view might have also been influenced by the implementation of a new decentralization policy, just before the reported intervention, which sought to increase the inclusion of varied stakeholder input, including that of the public (Byskov et al., 2014;Zulu et al., 2014).
Other actor life-world categories did not reflect a particular form of power as distinctly as ideological world-views. Nonetheless, they were useful to understand 'reactions' to the exercise of power by less powerful actors. For example, in several studies, we noted that the 'personal concerns' of DHMTs oriented them away from attention to public feedback. As reported by Tuba et al., the DHMT failed to act on complaints related to discrimination in waiting times and the provision of malaria supplies to politicians' relatives (Tuba et al., 2010). We judged this to be a power practice underpinned by the personal concerns of district health managers who feared that acting on these complaints would trigger workstation transfers instigated by politicians (Tuba et al., 2010). In the Tajikistan study, DHMT members were paid such low subsistence wages that most of their visits to facilities were focused on rent-seeking and punitive actions against frontline providers for 'wrong-doing' (Jacobs and Baez Camargo, 2020). This domination over frontline providers, coupled with contestation over resources between the DHMT and district hospital director and his team actions, was shaped by an interaction of 'power relationships and personal concerns' (reflected by low wages) and had an overall effect of undermining public trust in district officials.

Effects of power practices and forms of power on receiving and responding to public feedback
Drawing on the experiences highlighted in Table 6, responsiveness appeared to be enhanced by collaborative power practices while contestation, domination and resistance often undermined receiving and responding to public feedback. For example, in Tanzania and Zambia, collaborative power practices drawing on the technical expertise of the action research team and the positional power of the DHMT led to the opening up of the closed space within the DHMT. The authors reported greater inclusion of other DHMT members and stakeholders in the priority setting, creation of an opportunity for the public to appeal and revitalization of community participation structures to support the collection of public feedback Zulu et al., 2014;Blake et al., 2016).
In the social accountability intervention studies, NGOs drew on their resources, expertise and reputational power to support the functioning of feedback channels (Tuba et al., 2010;Blake et al., 2016;Boydell et al., 2020;George et al., 2018;Butler et al., 2020), while DHMTs drew on their positional power to address some public feedback. These collaborative power practices created a virtuous cycle that enhanced responsiveness. First, the generation of visible responses to public views, such as increased access to commodities (Tuba et al., 2010;Blake et al., 2016;Boydell et al., 2020), re-starting of services and infrastructural improvements in service delivery and addressing of kick-back practices at community and facility level (George et al., 2018;Butler et al., 2020), suggests that services were better aligned to community needs. Second, even where there was no immediate change in service delivery, responses such as DHMTs escalating issues to higher system levels or simply acknowledging community concerns increased the confidence of the public in voicing their needs. For example, in Malawi, a byproduct of well-performing community dialogue forums in an NGO-supported district was that community members set up forums in other districts without NGO support (Butler et al., 2020). Third, we identified reports of improved relationships between health providers and the public. For example, Blake et al. reported that an improved understanding of health providers' difficult working environment contributed to the creation of a midwife award system by a local traditional leader (Blake et al., 2016). In this study, one villager observed: Now I understand why they refer people. It's because they are not at the level where they can take care of certain problems. Previously I thought they were not ready to help us (p 376) (Blake et al., 2016). In contrast, domination, contestation and resistance, at the politicians/public interface at the district level, created a vicious cycle of low attendance that undermined the functioning of the public participation forum as a feedback channel and continued the marginalization of vulnerable groups (Kapiriri et al., 2003;McCollum et al., 2018). However, in the Ugandan study by Boydell et al., where the public also resisted working with politicians, domination by politicians was tempered by the presence of other feedback mechanisms 7 . In this study, the effects of contestation and resistance at the public/politician interface may also have been reduced by the district-level collaboration with an NGO that had adequate resources and the technical expertise to support VHTs (the feedback mechanism preferred by the public), which linked back to district health managers (Boydell et al., 2020). Finally, in the Tajikistan study (Jacobs and Baez Camargo, 2020), the public, aware of the power struggles at the district level, evolved an informal answerability mechanism directly with frontline providers and shared feedback directly with NGO service providers. Both processes did not link back to the DHMT, and thus, they were locked out of the process of receiving public feedback.

Discussion
This synthesis contributes to the literature on health system responsiveness by illuminating some of the actions taken by DHMTs in receiving and responding to feedback. However, the experiences reported in this synthesis have also highlighted weaknesses in the practice of responsiveness. These weaknesses included constraints to receiving feedback from the public (particularly vulnerable populations), little analysis (processing) at the DHMT level, inconsistent generation of responses and little communication to the public on generated responses. Few of the reviewed studies examined the role of power dynamics in DHMT responsiveness to public feedback in detail. Hence, we conducted a power analysis to understand our observations and generate ideas about how responsiveness might be strengthened. This synthesis adds to the emerging literature on responsiveness as a complex concept (Lodenstein et al., 2017;Mirzoev and Kane, 2017;Khan et al., 2021) pointing out the importance of actor interactions, power dynamics and varied elements of context as features of that complexity.
In the studies reviewed, DHMT members commonly exercised visible power linked to their managerial role. However, DHMT members were not uniformly empowered; some studies showed that core teams within the DHMT (closed spaces) had significant power linked to their access to resources and positions. The decisions and actions of these core teams influenced how the DHMT as a whole handled public feedback. We also noted that the public sector bureaucracy within which the DHMT operated held a form of invisible power embedded in its organizational culture that influenced to what extent DHMTs were willing and able to respond to public feedback. Several studies showed that politicians exercised hidden power, which influenced who was invited to share public feedback and what issues were included as priorities for discussion. Finally, we systematically identified invisible power as manifested in the subconscious influence of social norms, wider public governance, structures and discriminations that kept the public from providing feedback in the first place and that shaped the extent of responsiveness by DHMTs to public feedback.
Gaventa's power cube and Long's interface analysis were found to be complementary in analysis. The power cube supported the examination of the DHMT as a collective space and how this collective's use of power was supported or constrained by structural factors. We found these structural factors to be related to the power cube's levels of power and visible and invisible forms of power. For example, the national and regional levels of power commonly exercised visible power over DHMTs. Long's actor interface analysis was useful in eliciting where and with whom power lies within the DHMT and within the health system the DHMT is part of and why certain actions were taken (or not) by the DHMT concerning public feedback. Based on these findings, Figure 3 summarizes our ideas about the influence of power dynamics on district health managers' actions in receiving and responding to public feedback.
This framework illustrates how structural influences and the agency of actors interplay within the spaces where decision-making about public feedback happens. It suggests that actors' life-worlds are shaped by the contexts in which they find themselves. These in turn shape the actors' power practices and forms of power in receiving and responding to public feedback. Within a processing space for public feedback such as the DHMT, power can be wielded in both positive and negative ways. How this power is exercised has a reinforcing effect on the public's sharing of feedback. Positive power practices support the generation of responses and even more feedback from the public. Negative power practices could limit the generation of responses and the public's sharing of feedback or prevent the public from building claimed spaces. However, causation is not linear as actor interfaces form and re-form resulting in power struggles, the effect of which could be to support or undermine the practice of responsiveness, including by excluding the voices of marginalized groups. Furthermore, in these power struggles, power may flow bottom-up, contrasting with the traditional top-down flow, particularly where the public reacts to domination. These findings are relevant to HPSR investigators with an interest in health system responsiveness. They could, for example, build on this article and extend the framework presented in Figure 3 with research that considers experiences in other types of spaces such as HFCs or public participation forums where public feedback is received and responded to.
Our findings suggest that responsiveness might be strengthened by recognizing and building on actor life-worlds, while paying attention to the broader context in which the life-worlds are embedded. For example, politicians were observed to dominate the public and DHMTs, a power practice underpinned by the personal concerns of advancing political careers. In decentralized contexts such as Kenya (Nyikuri et al., 2017;Tsofa et al., 2017;McCollum et al., 2018) and Uganda (Kapiriri et al., 2003;Razavi et al., 2019;Boydell et al., 2020), the critical resource allocation and decisionmaking roles of political actors appeared to enhance this practice. Thus, leveraging the personal concerns of politicians (such as the interest to appeal to their voter base) in such contexts could deepen the practice of responsiveness to public feedback. The importance of recognizing the influence of political power in supporting policy implementation has been demonstrated in other published literature (Dalglish et al., 2015) although this study reported findings of a highly centralized political context. These findings are of value to health managers, particularly those who interact with political appointees and elected political representatives, as they draw attention to the need to appreciate the motivations of political actors who influence health system resourcing, planning and implementation.
Leveraging politicians' personal concerns requires careful application. This is because such an approach could direct responsiveness away from vulnerable groups (who often do not form a large voter base), thus undermining the equity goal of responsiveness. To address this challenge requires lowering the costs of participation in feedback channels. Health managers in collaboration with CSOs can lower participation costs by developing interventions aimed at off-setting invisible power by building the agency of the public, particularly vulnerable groups. Specific actions include increasing information available to the public regarding how their voices can be heard and supporting the public to present their concerns. In the reviewed articles, these activities were mainly conducted by NGOs, which raised the public's awareness about their rights and supported participatory platforms where citizens engaged with duty bearers at the community, health facility and district levels (Butler et al., 2020;Blake et al., 2016;George et al., 2018;Boydell et al., 2020;Butler et al., 2020). DHMTs can also participate in efforts to share power with the public by strengthening feedback channels, particularly those within the DHMT's mandate such as HFCs or VHTs. Such efforts could include vigilance to ensure that invited spaces are truly inclusive (including marginalized groups) and support participants' effective involvement. Specific actions here include, for example, providing timely information on invitations to public participation meetings and on-going rather than one-off engagement of the public (Shayo et al., 2012).
DHMTs' life-worlds related to their managerial positions of authority can also be leveraged to strengthen responsiveness. The study findings suggest that collaborative practices appear to hold promise for building responsive systems. Efforts by NGOs and research teams therefore need to support DHMTs to receive and respond to feedback, rather than working in parallel. Where these processes occur through a feedback channel not supported by the public health system, there needs to be a link back to the DHMT and public health system decision-makers. Such an approach could support learning and system-wide change. In the reviewed papers, where NGOs worked to strengthen pre-existing channels with the participation of DHMTs, there seemed to be increased trust in health system agents and improvements at the system level (Blake et al., 2016;George et al., 2018;Boydell et al., 2020;Butler et al., 2020). However, where NGOs operated independent of the DHMT and evolved their feedback mechanisms, such as in the study by Jacobs and Baez Camargo (2020), there was little reported improvement in public trust in district-level actors, and hardly any public feedback reached the DHMT.
Observations of organizational hierarchies in the reviewed articles suggest that regional and national health managers have the power to influence DHMTs to be more responsive to public feedback. Drawing on top-down implementation theory (Hill and Hupe, 2002), regional-and national-level actors could align resources and organizational environments to support receiving and responding to feedback at the district level. They could also hold DHMTs accountable for weak or no handling of public feedback. However, hierarchical power would need to be exercised to provide a supportive environment rather than demanding compliance. Literature cautions that multiple demands for compliance push managers to prioritize certain courses of action over others and that this could undermine responsiveness to the public (Nxumalo et al., 2018). In this review, many DHMTs experienced constraints on their flexibility to act due to guidelines and requirements for vertical performance accountability. To guard against this, emphasizing responsiveness to the public combined with transparency about actions taken in response to feedback and autonomy in decision-making is likely to contribute to orienting DHMTs outwards to the public and therefore to building responsiveness.
Another way to strengthen responsiveness could include efforts targeting at DHMTs world-views. In the studies reviewed, we identified mindsets among DHMT members such as little value for public feedback. Literature on strengthening district-level leadership and management suggests that setting up platforms for reflection and supportive supervision among DHMTs has the potential to shape mindsets about the value and legitimacy of public participation in health system decision-making (Cleary et al., 2017;Nzinga et al., 2021). Reflective practice can yield positive results in improving leadership and individual and team behaviours (Cleary et al., 2017;Nzinga et al., 2021). However, for reflective practice to have these effects, certain organizational conditions need to be in place that allows individual and group reflective practices to trigger organizational change (Nicolini et al., 2003). Nicolini et al. (2003) suggest that such an organizational change is possible even in highly fragmented and politicized organizations if the reflective practice is participatory and has the support or authorization of higher system levels (Nicolini et al., 2003). In LMIC contexts, this would include the support of regionaland national-level bureaucrats.
This review has some limitations. By including only English-language articles, we excluded several studies from Lusophone and Francophone Africa and Latin America that might have offered insights into the study questions. The inclusion of only English language articles might also explain why a majority of papers reported on experience in African countries. While the majority of papers mentioned decentralized study settings, there were not enough contextual data to determine the form of decentralization that is whether deconcentration, delegation or devolution (Mills et al., 1990) for all the studies. This paper is therefore limited in the extent to which it can draw conclusions on the differences in responsiveness across varying levels of decentralization. However, it is not unusual for syntheses to 'work with an incomplete knowledge base' (p 3) (Gilson, 2014) and, despite these limitations, our interpretive synthesis can provide a platform for future empirical work. Our synthesis work drew on a conceptual framework, which was both tested and adapted through this process. The adapted framework that we present therefore presents analytic generalizations of wider relevance.

Conclusion
In adopting an interpretive synthesis approach and applying two complementary power lenses, this work has systematically identified the influence of social norms, structures and discrimination on power distribution among actors in the environment surrounding, and within, the DHMT in relation to health system responsiveness. Furthermore, our analysis of power has illustrated reactions to the use of power and nontraditional flows of power (beyond the commonly reported top-down flows of power from national to regional to local and then to individual). The review has also proposed a conceptual framework (Figure 3) that can be applied to consider how receiving and responding to public feedback plays out in other health system spaces. The findings emphasize the need for measures that recognize the varied life-worlds of the range of actors involved in receiving and responding to public feedback. Some of these measures include leveraging politicians' power and personal interests while strengthening feedback channels to ensure meaningful public involvement and inclusivity, and interventions to shape DHMTs' world-views and work environments to support responsiveness to public feedback.

Supplementary data
Supplementary data are available at Health Policy and Planning online.