Decision space approach

One framework for analyzing decentralization is the “decision space” approach [13], which enables analysis of the amount of choice (i.e. wide, moderate, or narrow) transferred from higher to lower levels, the decisions made at local levels within this granted “space,” and the effects that these decisions have on the health system. Previously, decision space has been reported as mostly wide in the Philippines [14], mostly narrow in Ghana [14] and India [15], almost none in Fiji [16], and moderate or varying in Zambia [14], Uganda [14,17], and Pakistan [18,19]. The studies in Pakistan have suggested that a wide decision space at local levels may not be enough to improve service delivery unless it is accompanied by building the capacity of decision-makers who assume the new tasks and by ensuring accountability for the decisions they make. It is therefore useful for policy to explore effectiveness given multiple configurations of decision space with these dimensions of capacity and accountability [20]. We will contribute to this endeavor by an examination of health sector devolution in the Philippines.

Devolution in the Philippines

The Philippines is a republic in Southeast Asia comprising >7,600 islands and with a population of 101 million [21]. Government was historically highly-centralized through successive occupations by Spain, the United States, and Japan, followed by independence in 1946, and the dictatorship of Ferdinand Marcos that ended after a peaceful revolution in 1986. Since the 1950s, various waves of decentralization have taken place to disperse the concentration of power in Manila [22,23]. The largest wave culminated in the Local Government Code of 1991 [24] which introduced devolution to the entire archipelago in 1992, at that time considered as the most extensive decentralization in Asia [25]. With assistance from multi-lateral development organizations such as the World Bank, WHO, and USAID [22], the national government transferred the responsibility for government health services and other non-health services (e.g. agriculture) to Local Government Units (LGUs) across the archipelago [22–30]. Currently, the Philippines is organized into 17 regions for the purpose of coordination, although real political power at local levels lies with the LGUs that number into 81 provinces, and the 1,490 municipalities and 145 cities that are geographically within these provinces. Provincial governments maintain some oversight over municipalities and less-urbanized cities, while highly-urbanized cities are completely independent. Devolution disrupted the administrative structure of government health services from what used to be a service under a singular national ministry called the Department of Health (DOH), which previously managed an intact district health system at local levels, into a fragmented service under the control of individual LGUs: provinces responsible for hospitals; municipalities responsible for primary care facilities called Rural Health Units (RHUs); and cities responsible for both levels of care (Fig 1). Concepts such as “Interlocal Health Zones” [23] and “Service Delivery Networks” have been introduced by the national government to restore interlinking by fostering a network of health facilities and providers in contiguous areas, despite these facilities being under different LGUs, to offer a package of health services in an integrated and coordinated manner.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Simplified overview of the administrative structure of government health facilities in the Philippines before and after devolution.

https://doi.org/10.1371/journal.pone.0206809.g001

In the peer-reviewed literature, studies reported how devolution in the Philippines failed to enhance community participation in some municipalities [25] and sustained corruption when politicians became the center of decision-making [22] due to what has been described as “elite capture” [31] wherein existing power structures persist despite decentralization, compounded by a lack of accountability measures. In some provinces, inefficiency emerged as a problem when these took on more hospitals than what provincial resources could handle [22,26,28]. In some municipalities, patients learned to cross borders in search for better care [26], while a quality assurance program launched by the national government in 1998 failed to improve quality in primary care centers owned by the municipalities [32]. In the aspect of financing, municipalities, unlike the wealthier cities, continued to rely on the income from the national government for health spending [26]. Moreover, the lack of readiness at local levels prompted the national government to provide a training program in management for local decision-makers [30], and to deploy centrally-hired health professionals to municipalities that have no resources to hire them [33]. One paper on the malaria control program described poor implementation at local levels due to dysfunctional linking with the national level [27].

Consequently, we should then ask: What conditions enable decentralization to produce well-functioning health systems? [6]. We have explored this question by analyzing the perspectives of decision-makers at different levels of the Philippine health system. This is timely not only because of the 25 years of experience of implementing devolution in the Philippines, but also because of current initiatives in the country to change the structure of government from a republican into a federal state [34], indeed a step even further than devolution that will significantly alter how health services will be governed in the country. Lessons from the Philippines can offer policy-relevant insights [29] for countries that have decentralized or are contemplating to adopt some form of decentralization for their health systems.

Semi-structured questionnaire

A semi-structured questionnaire (S1 File) was developed by drawing from the decision space approach and the concept of health sector functions [13], as well as from two studies in Pakistan that analyzed the synergies between decision space, capacity, and accountability [18,19]. The questionnaire provided latitude in exploring participants’ insights and probed their perspectives on and personal experiences in implementing devolution. We examined their flexibility in making decisions within selected health sector functions. These functions were broad categories of tasks where decision-makers make choices for the health sector as previously reported in the studies by Bossert [13,14,18,19]. Drawing from these studies, we initially identified these functions as: 1) planning; 2) health budgeting and financing; 3) human resources for health management; and 4) service delivery.

Participant selection

We purposively-selected and contacted (via phone calls and emails) decision-makers who were serving the government health sector in positions of authority. Broadly, they represented three groups of decision-makers: 1) ministers and directors from the DOH who served at national and regional levels; 2) provincial, city, and municipal health officers, or those who served as career health officers at local levels; and 3) provincial governors, city mayors, and municipal mayors, or politicians who were elected to head the LGUs at local levels.

Data collection

The questionnaire was pilot-tested with two potential participants to check for clarity of the questions prior to use. One of the authors (HJL) with training in qualitative research conducted interviews face-to-face with each participant in his/her preferred venue in the Philippines between January and April 2017. HJL is also a Filipino citizen who is familiar with the country’s health system mostly through his work as an academic researcher and who was not employed in the government health service sector. Each interview was audio-recorded and manually transcribed in English. Transcripts were reviewed at least twice to ensure accuracy and subsequently loaded into MAXQDA Standard 12 (VERBI GmbH Berlin, 1995–2017) for coding and analysis.

Framework method

Analysis was based on the “Framework Method” as previously described in three papers [35–37]. It is considered a systematic approach to thematic analysis that compares and contrasts perspectives. Our approach to analysis combined both deductive and inductive approaches and is summarized as follows: 1) constant familiarization with the data through repeated listening to the audio-recordings while simultaneously reading the transcripts; 2) open coding of the transcripts that identified a preliminary set of categories based on the decision space and the health sector functions; 3) development of an initial analytical framework comprised of these codes and categories being identified from the transcripts; 4) coding of the rest of the transcripts using this analytical framework with continuing iteration whenever new categories were identified; and 5) analysis through comparison of emerging themes across categories, individual interviews, and groups of decision-makers with the use of tables.

Final thematic analysis focused on interpreting: 1) how decision space was exercised by the decision-makers in various health sector functions; 2) whether decision space was seen as wide, moderate, or narrow within each health sector function; and 3) the conditions that make decentralization effective for the health sector in the performance of these functions. We defined a condition as any factor or process (including any potential interaction between these) that has an enabling role in achieving a well-functioning decentralized or devolved health system. Similarly, we also identified those conditions that work in the opposite (i.e. hindering condition). We then summarized these enabling and hindering conditions in a table organized according to health sector functions, together with the decision space within these functions (using blue color coding) across groups of decision-makers. Finally, through an iterative process we synthesized the content of this table into a conceptual diagram, which was inspired by the image of decentralization and centralization previously described in the literature as movements between two opposite poles [2].

Ethics statement

Written informed consent was obtained from all participants prior to the conduct of the interviews. The study protocol was reviewed and approved in Switzerland by the Ethikkommission Nordwest- und Zentralschweiz (no. 2016-00738) and in the Philippines by the National Ethics Committee (no. 2016-013). Drafting of this paper was guided by the Consolidated Criteria for Reporting Qualitative Research (COREQ) [38].

Profiles of the decision-makers

We contacted 33 potential participants and interviewed up to 29 decision-makers when saturation was assessed to have already been achieved [39]. The audio files of two interviews were corrupted and subsequently excluded, which nevertheless did not change our judgment of saturation, resulting in a total of 27 interviews transcribed. Each interview lasted an average of one hour and four minutes. The 27 decision-makers worked in a wide range of local settings in the Philippines (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Present and previous areas of health sector-related work of the 27 decision-makers.

Locations indicate assignments that were ≥3 years. (Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.).

https://doi.org/10.1371/journal.pone.0206809.g002

There were 17 (63%) males and 10 (37%) females, with an average of 23.6 years of working in the Philippine government sector. At the time of the interviews, 10 (37%) were serving at national and regional levels, 11 (41%) were career health officers at local levels, and six (22%) were elected local officials. Many of them crossed different levels of government during the span of their careers. For instance, nine served in the DOH in various capacities, three of whom were once the Philippine Secretary of Health (i.e. Minister of Health). Among career health officers at local levels were four provincial health officers, three city health officers, and eight municipal health officers, four of whom were heads of their respective national associations of health officers. Among elected officials were three provincial governors, four municipal mayors, one city mayor, two congressmen, and one senator. We further characterized the length of service of each of the 27 decision-makers in Fig 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Durations of government service of the 27 decision-makers, the institutions they worked in, and their levels of decision-making.

Selected events in the Philippine health sector are also indicated.

https://doi.org/10.1371/journal.pone.0206809.g003

Health sector functions

The various decision-making activities described by the participants during the interviews led to an expansion of the initial list of health sector functions into five categories, namely: 1) planning; 2) financing and budget allocation; 3) resource management (further divided into “facilities, equipment, and supplies” and “human resources for health” or HRH); 4) program implementation and service delivery; and 5) monitoring and data management.

Planning

Devolution empowered LGUs to create the Local Health Board (LHB), a multi-stakeholder board chaired by the governor (in provinces) or mayor (in municipalities and cities) that serves as a venue for discussing local health concerns [25]. To what extent the LHB contributes to planning depends on whether it actually meets regularly, as the governor/mayor may choose not to convene it at all, and the ability of members to advocate for the concerns of the sectors they represent. After about 10 years since the introduction of devolution, the DOH instituted the annual “Investment Plan for Health” (IPH) [40] to assist the LGUs in planning and to restore some form of standard planning process. The IPH enables the DOH, through its regional offices, to train the LGUs to develop their annual plans for health, which specify local health needs and the resources from local and central levels to support these needs. Thus, the DOH has become actively involved in planning for local health services and is seen to have a wide (dark blue) decision space in this function compared to local decision-makers whose space may be described as moderate (blue) (Table 1). Conditions that enable decentralization in planning to be more effective for the health system include a functional LHB that feeds into the planning process, as well as opportunities for key decision-makers from central and local levels to meet, negotiate, set priorities, and co-create the local plans together. On the other hand, hindering conditions include a weak mechanism to monitor faithful execution of these plans, and the lack of sustainability for these plans given the reality of elections in the Philippines where local elected officials may change every three years.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Decision space at central and local levels for the functions of planning and financing and budget allocation (dark blue: Wide decision space; blue: Moderate; light blue: Narrow).

Enabling and hindering conditions are described.

https://doi.org/10.1371/journal.pone.0206809.t001

The following quote illustrates how planning has provided an opportunity for negotiation between the central and local levels and why it needs to be more strategic:

“Parts of the plans will be funded by the national government. We work with governors and mayors because the plans emerge from municipal and city levels and integrated at provincial level. The governor presents the consolidated plan and have it approved by the DOH regional office. That’s better because he’s the head of the province and will have ownership of the plan. But sometimes the plan is a wish list, for example, requesting the DOH to finance the fencing of their hospital [laughs]. Planning should be strategic to address real needs and improve their health system.”

(Director in the DOH central office, 28 years in government)

Financing and budget allocation

Most financing for health remained within the control of the national government, which pools tax collection and allocates the revenue share of LGUs based on a formula that considers local population and land area. It was the consensus among decision-makers that the inadequate share received by the LGUs led to the chronic underfunding and deterioration of many local health facilities especially in resource-poor provinces and municipalities that have little capacity for locally-generated income. Local health services often competed with other non-health services in budget allocation, which relied on the approval of the governor or mayor. In 1995, the Philippines created the Philippine Health Insurance Corporation (PhilHealth) [41], a DOH-attached agency that manages the national social health insurance program which is financed through premium contributions from enrolled members, most of whom come from the formal sector. Through PhilHealth’s reimbursements for services rendered, many local health facilities received additional financing to sustain operations. In financing and budget allocation, decision space is therefore seen as wide (dark blue) for the central level, and also wide (dark blue) for the local elected official who makes the decision on budget allocations, but moderate (blue) for the local health officer who, in most cases, needs the approval of the elected official when it comes to the local health budget. Enabling conditions include the institutional capacity of LGUs to raise revenues on their own, a well-funded central agency able to augment the lack of financial resources at local levels, and effective collaboration between the local elected official and health officer to be able to agree on allocating a substantial portion of the local budget in favor of health services. Hindering conditions include the concentration of financial resources at central levels despite devolution, and budget utilization that is driven mostly by political motivations [42] instead of genuine health needs.

The following quote illustrates an example of how central support helps the LGUs meet their needs in terms of financing and budget allocation:

“I kept talking to the municipalities to fix their RHUs. Of course, they have their allotment from taxes collected by the national government. But the important thing is for LGUs to understand that their operations are sustainable. How? They spend PHP100,000 (USD2,000) from their own budget to upgrade the RHU and have it accredited by PhilHealth as a maternal delivery unit. PhilHealth will pay PHP8,000 (USD160) for every delivery. How many deliveries, 30 per month? They get back PHP240,000 (USD4,800) per month. And that’s just for maternal health. The RHUs do many other things that PhilHealth will pay for.”

(Former Philippine secretary of health or “minister of health”)

Resource management

Despite devolution, the DOH continued to purchase the supplies needed for most public health programs, and these supplies are given to the LGUs as augmentation for their health facilities. In 2007, the DOH also initiated the “Health Facilities Enhancement Program” (HFEP) which provided a mechanism for LGUs to request assistance in the construction or upgrade of health facilities through funds from the national government. The DOH also established a national rural physician deployment program called “Doctors to the Barrios” [33] one year after the introduction of devolution which enabled the national government to hire physicians who are then deployed as local health officers in resource-poor municipalities that lack them. Deployment has since expanded to include nurses, midwives, medical technologists, and dentists. Under this program, deployed HRH receive their salaries from the national government but perform their duties as local HRH serving the LGUs. In some LGUs that have adequate resources to hire their own HRH, the governor or mayor has the supervisory authority over local HRH. Therefore, decision space for resource management overall is seen as wide (dark blue) for the central level, while at local levels it is moderate (blue) in terms of managing facilities, equipment, and supplies. However, for HRH management at local levels, decision space is seen as narrow (light blue) for the local health officer but wide (dark blue) for the local elected official who, in practice, is in full control of the hiring and firing process (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Decision space at central and local levels for the functions of resource management, further classified into facilities, equipment, and supplies and human resources for health (dark blue: Wide decision space; blue: Moderate; light blue: Narrow).

Enabling and hindering conditions are described.

https://doi.org/10.1371/journal.pone.0206809.t002

The following quote illustrates an example of how the devolution of HRH management led to inadequate compensation for local HRH especially in resource-poor LGUs:

“Public health workers are at the mercy of the LGUs in terms of salaries and benefits. The compensation enacted by the national government should also be given to local health workers. But the implementation of the standard salary rates is not the same across the country because the LGUs always say that they are autonomous from the national government. So the health workers in municipalities, cities, and provinces where the benefits are being given are lucky. But the others who don’t get these benefits still need to lobby for their rights.”

(Municipal health officer in a low-income island, 16 years in government)

On the other hand, the following quote illustrates the continuing significant intervention of the central government in providing for the various resources needed by the LGUs for better service delivery at local levels:

“There is creeping re-centralization in infrastructure, equipment, and human resource. The DOH also procures all commodities for most of the major public health programs. TB drugs and vaccines are entirely procured by the DOH and given to the LGUs, and the LGUs no longer need to buy anything. What else is devolved there? If you would look at the Philippine national health accounts, LGU expenditures for health are going down while the budget of the DOH is getting higher.”

(Philippine undersecretary of health or “deputy minister,” 28 years in government)

Some of the enabling conditions include: a governor or mayor who considers local health services as an important component of his/her administration and is supportive of the needs of local HRH; a local health officer who has good management skills, refrains from partisan politics, and is actively involved in the association of health officers who are able to use their influence as a group to assert their rights and privileges; and a DOH and PhilHealth with adequate resources to augment resource needs at local levels. On the other hand, one hindering condition, particularly in areas that host deployed HRH as augmentation for their lack of staff, is the potential tension between the local mayor, who is the head of the LGU, and the deployed HRH, who is technically an employee of the DOH. In such a situation, conflict sometimes arises because of the ambivalence in the lines of authority when the agency responsible for managing the decentralized service is different from the agency providing the salary of the staff tasked to deliver that service. Other hindering conditions include: weakened leverage in negotiating prices of supplies and equipment when devolution obliged LGUs to negotiate individually with suppliers in procuring what is needed at local levels; weak accountability for LGUs when these do not provide the full range of salaries and benefits that local HRH legally deserve; and the lack of a stepladder for local health officers to pursue their career aspirations as the devolved structure limits their opportunities for promotion within the LGU where they are employed.

Program implementation and service delivery

Devolution provided opportunities for LGUs to develop and implement local health programs that address their own unique needs, especially in settings with a culturally-sensitive context or where access to care is geographically-challenging. However, most LGUs still continued to rely on the DOH for technical assistance in the implementation of many public health programs (e.g. Expanded Program on Immunization, Family Planning Program, TB Control Program, Environmental Health Program, etc.), which are determined and planned at the national or central level and cascaded down through the DOH regional offices for implementation by the LGUs at local levels. Moreover, health facilities located in the same area may have limited means of effective cooperation between one another when these facilities are owned by different LGUs and only artificially linked through informal networks. In this context, decision space is viewed as wide (dark blue) for the central level and moderate (blue) at local levels for both the health officer and the elected official (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Decision space at central and local levels for the functions of program implementation and service delivery and monitoring and data management (dark blue: Wide decision space; blue: Moderate; light blue: Narrow).

Enabling and hindering conditions are described.

https://doi.org/10.1371/journal.pone.0206809.t003

The quote below illustrates an example from one province on how devolution has allowed the LGU to deliver health services that are suitable to the local context:

“We are indigenous peoples, and we have practices that are culturally-appropriate but may be frowned upon at the national level. While we advocate for facility-based deliveries, in geographically-isolated areas, mothers deliver in the house. When I was mayor, we provided training for the husbands because, in our culture, the person who delivers, aside from the midwife, is the traditional village birth attendant or the husband. So at least there is basic training for the husbands. That was our innovation. We also designed our local hospitals so that there are areas where the patient’s family can stay to have an atmosphere like home.”

(Former provincial governor and municipal mayor, 26 years in government)

Enabling conditions for decentralization to be effective include opportunities for innovation for local decision-makers to improve service delivery, as well as strong leadership on the part of central decision-makers to provide continuing technical guidance to the local levels for program implementation. Hindering conditions include a weak mechanism to ensure fidelity of program implementation at local levels [27], and weak interlinking between local health facilities owned by different LGUs but nevertheless located in the same catchment area, which has reduced opportunities for resource-sharing and a seamless patient referral scheme.

Monitoring and data management

The Field Health Services Information System (FHSIS) [43], which is managed by the DOH, contains the official health data of the Philippine government. With devolution, the seamless flow of data from local levels to regional and central levels to complete the FHSIS has become more challenging, especially with the loss of direct administration by the DOH over data reporting by LGUs. Nevertheless, efforts have been initiated to help facilitate data management by promoting the use of different electronic tools for data transmission from local levels. Thus, decision space for monitoring and data management for the central level is seen as moderate (blue), while decision space is wide (dark blue) for the health officer who controls data collection at local levels and narrow (light blue) for the elected official who has little involvement in performing this function.

The following quote is an illustration of how devolution has made it more difficult to harmonize the collection and pooling of health-related data at the national/central level:

“We try to publish the FHSIS final report every year. We are having a bit of difficulty, especially in some areas, in collecting the data. But with all the support that we are providing to the LGUs, it is easier to make them obey and submit their reports to us. Previously, we tried e-FHSIS and we gave computers to the LGUs, but they were not able to encode the data, and there were problems with connectivity. The final FHSIS report is usually 2–3 years delayed because it takes a long time to collect the data from all these LGUs.”

(DOH regional director, 34 years in government)

The conditions that either enable or hinder decentralization to improve the health system are several, and our exploration of decision-making within the five health sector functions provided a more organized way of capturing these conditions. Using the image of decentralization and centralization as a movement between two opposite poles [2], we have further synthesized these conditions in a conceptual diagram that mapped where these conditions should be considered in terms of performing the functions, and in terms of decision-making at central and at local levels (Fig 4). In this figure, we have also included some of the conditions in the broader context where decentralization is placed based on the experience in the Philippines. Some of these contextual conditions include an enabling political environment and a law that makes decentralization difficult to reverse, the supporting role played by multi-lateral/bilateral development organizations that provide technical assistance in implementing devolution, and the increasing population which, particularly in the Philippines, calls for a more efficient delivery of health care at local levels.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. A conceptual diagram inspired by the image of decentralization and centralization as movements between two opposite poles.

Various conditions to be considered for decentralization to be effective in improving the health system are proposed.

https://doi.org/10.1371/journal.pone.0206809.g004

Conditions related to the performance of the decentralized functions include clarity of roles for the various decision-makers in the system, a sense of ownership for the decisions they make, and the transfer of sufficient resources to support performance of these functions at local levels. Conditions at local levels include adequate capacities which, at the individual level, should include priority setting, innovation, systems thinking, and evidence-informed public health. Lastly, conditions at central levels include the ability to enforce an effective accountability mechanism, and to recover some of the advantages of centralization, such as in pooling of funds for more efficient financing, gaining leverage through bulk procurement of supplies on behalf of local health facilities nationwide, central augmentation of the needs at local levels especially to ensure equity, and enhancing cooperation between local health facilities in the same locality. The experience of health sector devolution in the Philippines suggests that decentralization can be implemented in policy but, in practice, some forms of re-centralization take place to make up for the inadequacies at local levels that took on the functions. Thus, one of the challenges in devolving the health sector is identifying the right combination of decentralized and centralized functions, even as the health system remains broadly decentralized, in order to achieve optimal health system performance.