Abstract
Psychological first aid (PFA) is a form of psychiatric emergency care: the humane, supportive early response offered to people in the immediate aftermath of a disaster or acute trauma, which MeSH files under descriptor D000088782. It is defined as much by what it is not — not psychotherapy, not debriefing, not treatment for a diagnosed disorder — as by what it is: supportive actions any trained helper can deliver to reduce distress and foster coping. This article traces PFA from crisis theory through the two frameworks that structure it — Hobfoll's five essential elements and the NCTSN's eight core actions — to the field's central tension: near-universal adoption resting on thin controlled evidence. A worked example quantifies why PFA is built to scale.
Keywords: psychological first aid, disaster response, five essential elements
What Psychological First Aid Is
Psychological first aid is a supportive, non-intrusive early response to people affected by a disaster, emergency, or acute traumatic event, intended to reduce initial distress, meet immediate practical needs, and foster short- and long-term adaptive functioning. In MeSH the descriptor (D000088782) sits under emergency services, psychiatric, placing PFA among the acute, crisis-facing branches of behavioral practice rather than among the scheduled psychotherapies. The World Health Organization's field guide frames it as help that is humane, supportive, and practical — assessing needs, ensuring basic safety and comfort, and connecting people to information and resources — delivered without pressing anyone to talk about what happened (#ref-who-2011).
The definition is sharpened by a series of deliberate exclusions. PFA is not psychological debriefing, the single-session ventilation of traumatic detail whose evidence turned out to be at best useless and at worst harmful; PFA was designed in part as the humane replacement for it. It is not a clinical treatment for post-traumatic stress disorder or any other diagnosis, and it does not require the recipient to have a disorder — most people exposed to disaster are distressed but resilient, and PFA supports that natural resilience rather than pathologizing it. And it is not something only a mental-health professional can provide: its actions are designed to be delivered by trained volunteers, first responders, teachers, and community members, which is the property that lets it scale to a whole affected population.
What unifies the intervention is a stance rather than a script. The PFA provider works from the affected person's own expressed needs, follows rather than leads, protects against further harm, and treats connection to practical resources and social support as itself therapeutic. The lineage runs back to Erich Lindemann's study of acute grief among survivors of the 1942 Cocoanut Grove fire, which founded the crisis-theory premise that a brief, well-timed, humane contact in the acute window can shape a person's longer trajectory (#ref-lindemann-1944). PFA is the operationalization of that premise for the mass-casualty and disaster setting.
The 2007 consensus review distilled the evidence on mass-trauma intervention into five targets any effective response should promote. They are the yardstick for a program's specific actions, not a script. Toggle an element to remove it and see how much of the evidence-supported target set remains covered.
The elements are evidence-supported targets, not a checklist to be marked off mechanically; the meter is a teaching device for their completeness, not a measured outcome. Computed locally, not stored.
The Five Essential Elements
The most influential conceptual foundation for PFA is not a manual but a consensus paper. In 2007 Hobfoll and a large group of trauma experts reviewed the empirical literature on what actually helps in the immediate and mid-term aftermath of mass trauma, and distilled it into five essential elements that any effective intervention should promote: a sense of safety, calming, a sense of self- and community efficacy, connectedness, and hope (#ref-hobfoll-2007). The claim is not that these five are a treatment, but that they are the evidence-supported targets — the intermediate states an intervention should move people toward, whatever its specific actions.
Each element rests on its own body of research. Promoting a sense of safety reduces the physiological and cognitive reactivity that sustains a threat response, which is why re-establishing actual and perceived safety comes first. Calming addresses the anxiety, hyperarousal, and numbing that can otherwise consolidate into a disorder, and it underwrites the practical emphasis on slowing breathing, orienting, and reducing exposure to distressing reminders. Self- and community efficacy — the belief that one's own actions, and one's group's, can produce desired outcomes — counters the helplessness that trauma induces and predicts better recovery. Connectedness captures the robust finding that social support is among the strongest protective factors after trauma, so restoring contact with family, friends, and social networks is a core aim. Hope — an expectancy that things can improve — protects against the pessimism and catastrophic outlook that predict poorer outcomes.
The five elements matter because they give PFA a rationale independent of any single manual. A given field guide's actions are justified to the extent that they promote safety, calming, efficacy, connectedness, and hope; a proposed intervention that promotes none of them has no empirical claim on the acute window. The framework thus serves as both a design principle for new programs and an evaluative yardstick for existing ones, and it is the reason the disparate PFA models share a recognizable common core despite differing in their specific steps.
| Element | What it targets | Why it matters |
|---|---|---|
| Safety | Actual and perceived safety. | Reduces the physiological and cognitive reactivity that sustains a threat response. |
| Calming | Anxiety, hyperarousal, and numbing. | Reduces the acute distress that can otherwise consolidate into a disorder. |
| Self- and community efficacy | Belief that one's own and one's group's actions can produce outcomes. | Counters the helplessness that trauma induces and predicts better recovery. |
| Connectedness | Contact with family, friends, and social networks. | Social support is among the strongest protective factors after trauma. |
| Hope | An expectancy that things can improve. | Protects against the pessimism and catastrophic outlook that predict poorer outcomes. |
The Core Actions
Where the five elements say what to promote, the operational manuals say what to do. The most widely used is the Psychological First Aid: Field Operations Guide, developed by the National Child Traumatic Stress Network and the National Center for PTSD, which organizes the intervention into eight core actions (#ref-brymer-2006). They are meant to be applied flexibly and in response to need, not as a fixed sequence: contact and engagement, safety and comfort, stabilization (of the emotionally overwhelmed), information gathering on current needs and concerns, practical assistance, connection with social supports, information on coping, and linkage with collaborative services. Vernberg and colleagues set out the rationale and the evidence-informed design behind this model, positioning it explicitly as an innovation in disaster mental health that replaced the discredited debriefing paradigm (#ref-vernberg-2008).
The eight actions map onto the five essential elements rather than duplicating them. Safety and comfort and stabilization serve safety and calming; connection with social supports serves connectedness; practical assistance and information on coping build self-efficacy; and linkage with collaborative services, by securing a path forward, supports hope. This mapping is what keeps the operational guide anchored to the evidence base: each action earns its place by advancing one or more of the empirically supported targets, and the demo below makes the correspondence explicit action by action.
The NCTSN Field Operations Guide organizes PFA into eight core actions, applied flexibly in response to need. Step through them to see what each action is and which essential element or elements it advances — the link that keeps the operational guide anchored to the evidence base.
The action-to-element mapping is illustrative of the design rationale: each action earns its place by advancing one or more evidence-supported targets. Actions are applied by need, not as a fixed sequence. Computed locally, not stored.
The design intent behind the eight actions is task-shifting. Each is described in concrete, teachable terms precisely so that a trained non-specialist can perform it, because the scale of a disaster far exceeds the supply of mental-health professionals. The WHO field guide makes the same move for low-resource and global settings, reducing PFA to the memorable triad look, listen, and link — observe for safety and urgent needs, listen without pressing, and link to support and information (#ref-who-2011). Both manuals trade the depth of a clinical intervention for the reach of one that ordinary helpers can deliver, which is the trade the worked example quantifies.
What the Evidence Shows
The defining tension of the field is that PFA is both widely endorsed and thinly evidenced. It is recommended by the WHO, the International Federation of Red Cross and Red Crescent Societies, and national disaster-response systems, and it has been disseminated to hundreds of thousands of responders. Yet the direct evidence that it improves mental-health outcomes is strikingly limited, and the field has been unusually candid about the gap.
Fox and colleagues conducted an early analysis of the peer-reviewed literature from 1990 to 2010 and found that, although PFA rested on sound theoretical and expert-consensus foundations, there was almost no direct empirical evidence of its effectiveness as a disaster-intervention tool (#ref-fox-2012). Dieltjens and colleagues, in a systematic literature search, reached the same conclusion more starkly: they could find no controlled studies demonstrating that PFA works, and therefore no evidence base from which to build practice guidelines (#ref-dieltjens-2014). Shultz and Forbes framed the situation as a paradox — a rapid proliferation of PFA training and adoption running far ahead of the search for evidence to justify it (#ref-shultz-2014).
The reasons for the gap are partly practical and partly ethical. Disasters are unpredictable, chaotic, and ethically fraught settings for randomized trials: it is difficult and often unacceptable to withhold supportive contact from a control group in the acute aftermath of catastrophe. The result is that the evidence for PFA is largely indirect — inherited from the five essential elements, whose components (social support, a sense of safety, efficacy) are individually well supported — rather than direct evidence that the packaged intervention changes outcomes. Whether that indirect warrant is sufficient, given how widely PFA is deployed, is the question the recent literature has tried to answer.
Training and Scalability
Because PFA is delivered by trained non-specialists, the more tractable evidence question has been not does PFA help recipients but can people be trained to deliver it well. Here the findings are more encouraging. McCabe and colleagues developed a consensus-derived, competency-based training model — the Johns Hopkins RAPID PFA model — that specifies the observable competencies a provider should acquire and assesses them, moving PFA training from vague exposure toward a measurable curriculum (#ref-mccabe-2014). Defining the competencies is what makes training evaluable at all: a course can be judged by whether trainees actually acquire them.
The training studies that follow this logic report gains in knowledge, skills, and confidence. Wang and colleagues, in a scoping review of PFA training, found that across varied populations and delivery formats training reliably improved participants' PFA knowledge, self-efficacy, and skills, even though the downstream effect on affected people remained largely unmeasured (#ref-wang-2021). The scalability that motivates PFA thus has genuine empirical support at the level of the provider: ordinary helpers can be taught the competencies, retain them, and feel more able to help. The worked example makes the scaling logic concrete.
PFA is built out of teachable actions so a few experts can seed a response that reaches a whole population. A handful of master trainers train trainers, who train responders, who each assist affected people. Vary each tier to see the total reach and how it compares with the experts delivering PFA directly.
With the defaults (5 × 12 × 30 × 20), the cascade reaches 36,000 people — 360 times the 100 the 5 experts could assist directly. The multiplier is the rationale for a teachable intervention, and also its risk: reach multiplies whatever the true per-person effect is. Computed locally, not stored.
Worked Example
The reason PFA is designed for delivery by non-specialists becomes quantitative in a train-the-trainer cascade, the mechanism by which a handful of experts can seed a response that reaches a whole affected population. Suppose a disaster-response agency begins with 5 master trainers. Each master trainer trains 12 local trainers, giving
5 × 12 = 60 trainers.
Each of those trainers in turn trains 30 community responders — volunteers, teachers, and first responders — producing
60 × 30 = 1,800 responders.
Finally, suppose each responder provides PFA to 20 affected people over the course of a deployment. The total reach is
1,800 × 20 = 36,000 people,
or, as a single product, 5 × 12 × 30 × 20 = 36,000.
The contrast that makes the cascade worthwhile is the direct alternative. Had the 5 experts instead each personally assisted 20 people, the response would have reached 5 × 20 = 100 people. The cascade reaches 36,000 / 100 = 360 times as many. The multiplier is the entire rationale for building an intervention out of teachable, competency-defined actions rather than clinical expertise: the scarce resource — the experts' time — is spent on training rather than on direct delivery, and the reach compounds at each tier. The demo above lets the number of trainers, responders, and people at each level be varied, so the reach and the multiplier over direct delivery can be read off any staffing configuration. The same arithmetic also exposes the field's evidentiary risk: a mechanism that multiplies reach multiplies whatever the intervention's true per-person effect is — large if PFA works, and merely a large deployment of an unproven method if it does not.
Discussion
Psychological first aid occupies an unusual position in evidence-based practice: it is a reasonable default that has outrun its own evidence. The case for using it does not rest on trials showing it works, because those trials barely exist; it rests on three weaker but genuine warrants. First, the alternative it replaced — psychological debriefing — was actively harmful, so a supportive, non-intrusive stance is a safer default even absent proof of benefit. Second, its targets are borrowed from the five essential elements, whose components are individually well supported (#ref-hobfoll-2007). Third, it is low-cost, low-risk, and humane, which lowers the evidentiary bar a cautious field is willing to accept for deployment. The honest description is that PFA is plausible and prudent rather than proven.
This has made PFA a recurring case study in how a field should behave when adoption outpaces evidence. The candor of its own reviewers is notable: rather than defend PFA with weak studies, figures within the field have repeatedly published the gap itself (#ref-fox-2012; #ref-dieltjens-2014; #ref-shultz-2014). That candor is the appropriate response to the worked example's warning — because the cascade multiplies reach regardless of efficacy, the responsibility to establish whether the per-person effect is real grows with every responder trained, not shrinks. The field's task is to convert an intervention that is widely deployed on prudential grounds into one that is deployed on demonstrated ones, without abandoning the humane default in the interim.
Current Directions
The most direct attack on the evidence gap is the small but growing set of controlled studies of packaged PFA. Despeaux and colleagues conducted a randomized controlled trial of a group-based PFA protocol and found measurable effects on participants' distress, one of the first randomized tests of the intervention rather than of training alone (#ref-despeaux-2019). It is a single trial with the limits of its setting, but it marks the shift the field's critics called for: from documenting the absence of evidence to generating it.
A second front is the rigorous study of training itself, on the premise that a scalable intervention must first be shown to be learnable before its downstream effects can be isolated. Lee and colleagues evaluated a didactic-plus-simulation PFA program for students and school counselors in South Korea and found that the addition of simulation improved trainees' competence beyond didactic instruction alone, sharpening the question of how — not merely whether — PFA is best taught (#ref-lee-2017). Wang and colleagues' scoping review synthesized this literature and mapped where training research is strong (knowledge, self-efficacy) and where it remains thin (retention, and the effect on the people ultimately served) (#ref-wang-2021).
The most recent synthesis takes the measure of the whole enterprise. Hermosilla and colleagues conducted a systematic review of PFA's effect on mental health and well-being and concluded, in effect, that the field still needs to build the evidence — that the accumulated controlled data, even now, cannot firmly establish benefit or its absence, and that well-designed studies remain the priority (#ref-hermosilla-2023). The open questions are therefore not conceptual but empirical: whether packaged PFA changes outcomes for recipients, which components carry any effect, how best to train and retain providers, and how to study an intervention delivered in exactly the chaotic settings that make rigorous study hardest.
Common Misconceptions
- Psychological first aid is a form of therapy or counseling.
- PFA is a supportive, practical early response, not a clinical treatment. It does not require a diagnosis, does not treat a disorder, and is designed to be delivered by trained non-specialists, not only by mental-health professionals (#ref-who-2011; #ref-vernberg-2008).
- PFA is the same as critical-incident debriefing.
- It is close to the opposite. Debriefing pressed survivors to recount traumatic detail in a single session and proved ineffective or harmful; PFA was designed as the humane, non-intrusive replacement and never presses anyone to talk about what happened (#ref-vernberg-2008).
- Because PFA is so widely used, it must be proven to work.
- Wide adoption and strong evidence are not the same thing. Systematic reviews have repeatedly found little direct controlled evidence that PFA improves outcomes; its warrant is largely indirect and prudential, and building the direct evidence remains an active task (#ref-fox-2012; #ref-dieltjens-2014; #ref-hermosilla-2023).
- Everyone exposed to a disaster needs psychological first aid.
- Most people exposed to disaster are distressed but resilient and recover without intervention. PFA supports natural resilience and connects people to resources; it does not assume pathology, and it is offered in response to need rather than imposed on all (#ref-hobfoll-2007).
Glossary
- Calming.
- The second of Hobfoll's five essential elements: reducing the anxiety, hyperarousal, and numbing that can otherwise consolidate into a disorder after trauma.
- Connectedness.
- The fourth essential element: restoring contact with family, friends, and social networks, reflecting that social support is among the strongest protective factors after trauma.
- Core actions.
- The eight operational steps of the NCTSN Field Operations Guide — contact and engagement, safety and comfort, stabilization, information gathering, practical assistance, connection with social supports, information on coping, and linkage with services.
- Crisis theory.
- The framework, founded by Lindemann's study of acute grief, holding that a brief, well-timed, humane contact in the acute window can shape a person's longer recovery; the intellectual root of PFA.
- Debriefing (psychological).
- A single-session ventilation of traumatic detail once used after disasters; found ineffective or harmful, and the practice PFA was designed to replace.
- Five essential elements.
- Hobfoll and colleagues' 2007 consensus set of evidence-supported targets for mass-trauma intervention: safety, calming, self- and community efficacy, connectedness, and hope.
- Hope.
- The fifth essential element: an expectancy that things can improve, protecting against the pessimism and catastrophic outlook that predict poorer post-trauma outcomes.
- Look, listen, and link.
- The WHO field guide's memorable triad reducing PFA to its essentials: observe for safety and urgent needs, listen without pressing, and link to support and information.
- Psychological first aid (PFA).
- A supportive, non-intrusive early response to people affected by disaster or acute trauma, intended to reduce initial distress and foster adaptive functioning, deliverable by trained non-specialists.
- RAPID model.
- The Johns Hopkins consensus-derived, competency-based PFA training model developed by McCabe and colleagues, specifying and assessing the observable competencies a provider should acquire.
- Resilience.
- The capacity of most people exposed to disaster to remain distressed but functional and to recover without intervention; the natural process PFA supports rather than replaces.
- Safety.
- The first essential element: re-establishing actual and perceived safety, which reduces the physiological and cognitive reactivity that sustains a threat response.
- Self- and community efficacy.
- The third essential element: the belief that one's own actions and one's group's can produce desired outcomes, countering the helplessness that trauma induces.
- Task-shifting.
- Designing an intervention out of concrete, teachable actions so that trained non-specialists can deliver it, allowing the response to scale to a whole affected population.
- Train-the-trainer cascade.
- A tiered dissemination model in which master trainers train trainers who train responders, multiplying the reach of a small number of experts across successive levels.
Key Researchers
Melissa J. Brymer (living). Director of Terrorism and Disaster Programs at the UCLA-Duke National Center for Child Traumatic Stress and lead developer of the NCTSN/National Center for PTSD Psychological First Aid: Field Operations Guide. ORCID - Faculty
George S. Everly (b. 1950). Co-founder of the field of disaster mental health and psychological crisis intervention; co-author of the Johns Hopkins RAPID competency-based training model and of a randomized trial of group PFA. ORCID - Google Scholar - Faculty - Wikipedia
Stevan E. Hobfoll (b. 1951). Author of Conservation of Resources theory and lead author of the 2007 consensus paper defining the five essential elements of mass-trauma intervention that underpin PFA. Google Scholar
Erich Lindemann (1900-1974). His 1944 study of acute grief among survivors of the 1942 Cocoanut Grove fire founded crisis theory and the immediate-support tradition from which psychological first aid descends. Wikipedia - Wikidata
Josef I. Ruzek (living). Former director of dissemination and training at the National Center for PTSD; co-author of the PFA Field Operations Guide and of foundational early-intervention research. ORCID - Google Scholar - Faculty - Wikipedia
Patricia J. Watson (living). Senior educational specialist at the National Center for PTSD and co-developer of PFA and Skills for Psychological Recovery, and a co-author of the recent systematic review of the evidence. Google Scholar - Faculty
Frequently Asked Questions
What is psychological first aid? It is a supportive, non-intrusive early response to people affected by a disaster or acute traumatic event, intended to reduce initial distress, meet immediate practical needs, and foster adaptive functioning. In MeSH it is descriptor D000088782, filed under emergency psychiatric services.
Is psychological first aid a kind of therapy? No. PFA is not a clinical treatment and does not require a diagnosis. It supports the natural resilience of people in the acute aftermath of trauma and connects them to resources, and it is designed to be delivered by trained non-specialists rather than only by mental-health professionals.
How is PFA different from debriefing? It is close to the opposite of debriefing. Critical-incident debriefing pressed survivors to recount traumatic detail in a single session and was found ineffective or harmful; PFA was designed as the humane replacement and never presses anyone to talk about what happened.
What are the five essential elements? They are the evidence-supported targets Hobfoll and colleagues distilled in 2007: a sense of safety, calming, a sense of self- and community efficacy, connectedness, and hope. They function as the rationale and yardstick for PFA's specific actions.
What are the eight core actions? The NCTSN Field Operations Guide organizes PFA into eight steps applied flexibly in response to need: contact and engagement, safety and comfort, stabilization, information gathering, practical assistance, connection with social supports, information on coping, and linkage with collaborative services.
Does psychological first aid actually work? The honest answer is that the direct controlled evidence is limited. PFA is widely endorsed and its targets are individually well supported, but systematic reviews have repeatedly found little direct evidence that the packaged intervention improves outcomes, and building that evidence is an active area of research.
Who can deliver PFA? By design, any trained helper, not only clinicians: volunteers, first responders, teachers, and community members. This is what allows the response to scale to a whole affected population through train-the-trainer dissemination.
Can people be trained to deliver PFA well? Yes. Training studies reliably show gains in PFA knowledge, self-efficacy, and skills across varied populations, and competency-based models such as Johns Hopkins RAPID make the training measurable, even though the downstream effect on recipients remains under-studied.
References
Brymer, M., Jacobs, A., Layne, C., Pynoos, R., Ruzek, J., Steinberg, A., Vernberg, E., & Watson, P. (2006). Psychological first aid: Field operations guide (2nd ed.). National Child Traumatic Stress Network & National Center for PTSD. OCLC 705260763.
Despeaux, K. E., Lating, J. M., Everly, G. S., Jr., Sherman, M. F., & Kirkhart, M. W. (2019). A randomized controlled trial assessing the efficacy of group psychological first aid. The Journal of Nervous and Mental Disease, 207(8), 626-632. https://doi.org/10.1097/NMD.0000000000001029
Dieltjens, T., Moonens, I., Van Praet, K., De Buck, E., & Vandekerckhove, P. (2014). A systematic literature search on psychological first aid: Lack of evidence to develop guidelines. PLOS ONE, 9(12), e114714. https://doi.org/10.1371/journal.pone.0114714
Fox, J. H., Burkle, F. M., Jr., Bass, J., Pia, F. A., Epstein, J. L., & Markenson, D. (2012). The effectiveness of psychological first aid as a disaster intervention tool: Research analysis of peer-reviewed literature from 1990-2010. Disaster Medicine and Public Health Preparedness, 6(3), 247-252. https://doi.org/10.1001/dmp.2012.39
Hermosilla, S., Forthal, S., Sadowska, K., Magill, E. B., Watson, P., & Pike, K. M. (2023). We need to build the evidence: A systematic review of psychological first aid on mental health and well-being. Journal of Traumatic Stress, 36(1), 5-16. https://doi.org/10.1002/jts.22888
Hobfoll, S. E., Watson, P., Bell, C. C., Bryant, R. A., Brymer, M. J., Friedman, M. J., Friedman, M., Gersons, B. P. R., de Jong, J. T. V. M., Layne, C. M., Maguen, S., Neria, Y., Norwood, A. E., Pynoos, R. S., Reissman, D., Ruzek, J. I., Shalev, A. Y., Solomon, Z., Steinberg, A. M., & Ursano, R. J. (2007). Five essential elements of immediate and mid-term mass trauma intervention: Empirical evidence. Psychiatry: Interpersonal and Biological Processes, 70(4), 283-315. https://doi.org/10.1521/psyc.2007.70.4.283
Lee, J. S., You, S., Choi, Y. K., Youn, H. Y., & Shin, H. S. (2017). A preliminary evaluation of the training effects of a didactic and simulation-based psychological first aid program in students and school counselors in South Korea. PLOS ONE, 12(7), e0181271. https://doi.org/10.1371/journal.pone.0181271
Lindemann, E. (1944). Symptomatology and management of acute grief. American Journal of Psychiatry, 101(2), 141-148. https://doi.org/10.1176/ajp.101.2.141
McCabe, O. L., Everly, G. S., Jr., Brown, L. M., Wendelboe, A. M., Abd Hamid, N. H., Tallchief, V. L., & Links, J. M. (2014). Psychological first aid: A consensus-derived, empirically supported, competency-based training model. American Journal of Public Health, 104(4), 621-628. https://doi.org/10.2105/AJPH.2013.301219
Shultz, J. M., & Forbes, D. (2014). Psychological first aid: Rapid proliferation and the search for evidence. Disaster Health, 2(1), 3-12. https://doi.org/10.4161/dish.26006
Vernberg, E. M., Steinberg, A. M., Jacobs, A. K., Brymer, M. J., Watson, P. J., Osofsky, J. D., Layne, C. M., Pynoos, R. S., & Ruzek, J. I. (2008). Innovations in disaster mental health: Psychological first aid. Professional Psychology: Research and Practice, 39(4), 381-388. https://doi.org/10.1037/a0012663
Wang, L., Norman, I., Xiao, T., Li, Y., & Leamy, M. (2021). Psychological first aid training: A scoping review of its application, outcomes and implementation. International Journal of Environmental Research and Public Health, 18(9), 4594. https://doi.org/10.3390/ijerph18094594
World Health Organization, War Trauma Foundation, & World Vision International. (2011). Psychological first aid: Guide for field workers. World Health Organization. ISBN 9789241548205.