Evidence-based medicine is widely accepted to guide medical recommendations: clinical or diagnostic guidelines are developed on the grounds of the best available data, systematic reviews and sound clinical understanding (Sackett et al. 1996). But occasionally the quality of the evidence may not be as good as expected (Kett et al. 2011). In the absence of high-quality data, one has to rely on information generated through studies with suboptimal designs and based on mere empirical – frequently subjective – experience. Important medical problems occurring in high-income countries can, depending on site-specific conditions, provoke much more severe challenges when occurring in low- and middle-income countries (LMICs).
Most clinical guidelines not directly related to tropical medicine cannot or do not consider trials performed in LMICs, because, for many medical conditions, such trials have either not been conducted there or are of questionable quality. Technical prerequisites, ethical issues, infrastructural constraints, overburdened health systems and, frequently, the lack of appropriate funding prevent successful and meaningful realization of clinical trials in LMICs. As a result, conclusions on the applicability of guidelines that have proven valuable in high-income countries are difficult to draw for LMICs.
The clinical guideline for haemodynamic support of paediatric and neonatal septic shock proposed by the American College of Critical Care Medicine (Brierley et al. 2009), for example, suggests early fluid administration as a primary therapeutic measure. And, if applicable, this measure should be taken in conjunction with interconnected measures such as inotropic support, close haemodynamic monitoring and mechanical ventilation. This guideline and the algorithm proposed are largely accepted in high-income countries.
Severe febrile conditions causing a systemic inflammatory response syndrome are important causes of morbidity and mortality in LMICs. This applies particularly to children and fevers with underlying causes such as malaria and septicaemia. In LMICs, diagnostic resources and accurate clinical monitoring, including treatment options, are often limited. Recently a multicentre, open, randomized controlled trial on fluid management in Kenyan, Tanzanian and Ugandan children was designed to apply fluid boluses as suggested by the aforementioned guideline (Maitland et al. 2011). The children were suffering from severe febrile illnesses complicated by impaired consciousness and/or respiratory distress and showed evidence of impaired perfusion. Upon admission to hospital, they were treated with boluses of either albumin or saline. In addition, the children received maintenance fluids and, in cases of haemoglobin levels below 5 g/dl, whole blood transfusions of 20 ml/kg over a period of 4 h. Bolus volumes were increased after a protocol amendment. The result of the study was that fluid boluses significantly raised the 48-h mortality, and as a consequence, the trial had to be stopped.
The study was performed in ‘typical African hospitals, which have no intensive care facilities’ (Maitland et al. 2011). Although complications due to fluid overload were reported to be rare, the outcome of the study and its halting ahead of schedule recommended by the Data and Safety Monitoring Committee indicate that potential complications were not perceived in due time, most likely due to the lack of diagnostic means and merely clinical assessments. The investigators singled out and evaluated one specific intervention from the set of complex and interconnected procedures outlined in the guideline. This important study raises the fundamental question about the applicability of guidelines in LMICs when data were generated in high-income settings and vice versa, in addition to a number of weighty arguments put forward by others (Southall & Samuels 2011).
Chinnock et al. (2005) have emphasized that ‘the relevance of systematic reviews to frontline health care workers in developing countries has so far been limited’. This applies, in our view, to clinical guidelines as well. They may not always be applicable in LMICs, as they have not been developed for and in these settings. Recommended procedures may be inappropriate or even harmful, as the example of possible fluid overload in the absence of intensive care facilities has demonstrated.
Groups of physicians, scientists and organizations such the Conference on Guideline Standardization (COGS) or the New Zealand Guideline Group, at times supported by WHO, are concerned with the development and implementation of medical guidelines. Other groups try to modify and adjust existing guidelines according to the specific conditions and requirements in LMICs. For example, with regard to cardiovascular risks, different approaches have been developed by WHO for high-income countries and LMICs to prevent complications and undue expenditure (Mendis et al. 2011). The lack of sepsis guidelines for LMICs has been stated repeatedly (Jacob et al. 2011; Landre-Peigne et al. 2011). In 2007, the American Cancer Society and WHO presented resource level-appropriate guidelines for the overall management of major cancers in LMICs, which differ considerably from strategies applied in high-income countries (Anderson et al. 2008). Notably, and in addition to strict medical aspects, the Breast Health Global Initiative emphasized the need to develop evidence-based, affordable and culturally acceptable guidelines for use in LMICs. Numerous examples exist on the pertinent customization of surgical and gynaecological techniques in LMICs (e.g. Abulkhair et al. 2010; Nakahara et al. 2011, Osen et al. 2011). Guidelines and treatment strategies for the major infectious diseases HIV/AIDS, tuberculosis and malaria vary considerably between high-income countries and most LMICs.
The COGS suggests that guidelines should ‘describe the intended users of the guideline (e.g. provider types, patients) and the settings in which the guideline is intended to be used’ (Shiffman et al. 2003; Schünemann et al. 2006). Decisions made in the course of local adaptation processes, however, have to be as transparent as the initial process of guideline development (Schünemann et al. 2006). Meticulous guideline evaluation in LMICs applying powerful study designs should be mandatory. Simple adaptation without evaluation by use of adequate study designs is not good enough.
During site-specific adaptation, a guideline might lose parts of its evidence base and may, when locally applied, be less efficient or even harmful. Thus, we believe that more thought must be devoted to:
- • The setting in which a guideline has been developed and evaluated;
- • The decision processes addressing the question whether or not a guideline developed in high-income countries may also be suitable for LMICs;
- • Local adaptation of guidelines, and timely inclusion of relevant stakeholders and experts; and
- • Local evaluation of modified guidelines through either cost-effective implementation research or rigorous clinical trials.