Decentralisation , Decision Space and Directions for Future Research Comment on “ Decentralisation of Health Services in Fiji : A Decision Space Analysis ”

Decentralisation continues to re-appear in health system reform across the world. Evaluation of these reforms reveals how research on decentralisation continues to evolve. In this paper, we examine the theoretical foundations and empirical references which underpin current approaches to studying decentralisation in health systems.


Introduction
As a perennial concept in health policy and management, decentralisation remains an area of academic inquiry in many countries. Whilst empirical study remains the focus, conceptual investigations are generally less well-developed in the health sector. One of the most common conceptual lenses through which to view decentralisation is Bossert's notion of decision space. 1 In this paper (Mohammed et al 2 ), the decentralisation of health services in Fiji has been evaluated using the decision space framework of Bossert. 1 The case-study is decentralisation of health services from Ministry of Health (MoH) to the Health Centres along five main areas, including finance, service organisation, human resources, access rules, and governance rules. The main type of decentralisation in this context is de-concentration of outpatient services from tertiary hospitals at the divisional level to the sub-divisional health centres. However, health centres as the local agents do not have the authority to make decisions regarding finances, human resources, and governance rules. They only have very limited decision space over the access rules (targeting of the health services) and service organisation (in terms of offering services above the basic package).

Theoretical Foundation
This study applies the widely applied principal-agent approach; the MoH is the principal while hospitals at the divisional (tertiary and secondary) and sub-divisional (secondary and primary) level as well as health centres are the agents. The Bossert framework helps the authors to analyse the extent of vertical autonomy granted to health centres from the higher authority. This framework is appropriate and has been applied quite extensively especially in studies from low-and middleincome countries. The authors acknowledge the shortcomings of Bossert framework to capture the actual range of decisions that the agent may have at the local level. However, the scope of the literature is limited to the vertical decentralisation. The authors have not taken an alternative approach to explain the contextual factors such as organisational characteristics, financial position, and organisational capacity that can influence the decision space of the agents at the local level. The authors have made reference to the role of the commissioning bodies and the Ministry of Finance, but do not elaborate. Also, according to this paper, the MoH is both the purchaser and provider of health services. As elsewhere, this centre's motives and decision are significant factors in all health systems and merits further attention.

Empirical References
This study could benefit from presenting more empirical evidence. The only table of the data provided is the workload of health centres before and after decentralisation. It would be more illuminating if there was a table to show the sources of the data, the information that data gives and the limitations of the sources. Moreover, it seems that the data has only been collected from the MoH. This data should be corroborated by data of the health centres. Details about data analysis are also missing. The limitations of the study need to be stated clearly. More longitudinal studies of decentralisation (beyond a "before and after" design) are warranted, in order to assess the long-term effect of decentralisation (and centralisation). Equally, more extensive data of the impact of decentralisation across the entire health system is required to make a fuller assessment of such policy initiatives.

Conclusion
Overall, this study by Mohammed and colleagues gives a clear picture of the extent of the vertical decentralisation in Fiji's healthcare system. It would be insightful to use the relevant literature (eg, Exworthy and Frosini 3 ) to give the readers an overview about the characteristics of the health centres as local agent to see whether these centres have the required capacity to make decisions over different aspects of service delivery. This study adds to the existing evidence about decentralisation, and contributes to the conceptual development in this field. Given the enduring appeal of decentralisation to health systems across the world, research remains vital in reaching a better understanding of the scope and consequences of decentralisation programmes. Longitudinal and system-wide data would strengthen the evidence base considerably.

Ethical issues
Not applicable.