S. Y. Yu

Bio: S. Y. Yu is an academic researcher. The author has contributed to research in topics: Medicine & Internal medicine. The author has an hindex of 1, co-authored 1 publications receiving 5925 citations.

The International Classification of Headache Disorders, 3rd edition (beta version)

Abstract: The International Classification of Headache Disorders, 3 edition (beta version), may be reproduced freely for scientific, educational or clinical uses by institutions, societies or individuals. Otherwise, copyright belongs exclusively to the International Headache Society. Reproduction of any part or parts in any manner for commercial uses requires the Society’s permission, which will be granted on payment of a fee. Please contact the publisher at the address below. International Headache Society 2013. Applications for copyright permissions should be submitted to Sage Publications Ltd, 1 Oliver’s Yard, 55 City Road, London EC1Y 1SP, United Kingdom (tel: þ44 (0) 20 7324 8500; fax: þ44 (0) 207 324 8600) (www.sagepub.co.uk). Translations

Incidence, clinical correlates and associated outcomes of dementia in heart failure: a population-based cohort study

Abstract: Type of funding sources: None. Dementia, in the setting of heart failure (HF), portends poorer outcomes and poses great challenges in its clinical management. We investigated the incidence, types, clinical correlates, and the prognostic impact of dementia in a population-based cohort of patients with HF. Further, we examined the interactions of age and sex, and education status with dementia incidence. The previously validated Hong Kong Clinical Data Analysis Reporting System (CDARS), a territory-wide database was interrogated to identify patients with HF (N= 202,121) from 1995 to 2018. Associations of clinical correlates with incident dementia and its risk with all-cause mortality were assessed using competing risk/multivariable Cox regression models where appropriate. Among a total cohort aged ≥18 years with HF (mean age: 75.3 ± 13.0 years, 51.3% women), new-onset dementia occurred in 22,145 (11.0%) over a median follow-up of 5.5 years. Alzheimer’s disease occurred in 27.0%; vascular dementia (18.1%) and unspecified dementia (in 55.1%). Age-standardized rate of dementia incidence in women was 1297 (95%CI, 1276-1318) (vs. 744, 95%CI, 723-765) per 10000 population in men. Other independent predictors of dementia include: Increasing age (HR 1.08), Female sex (HR 1.19), Nil/< primary (vs tertiary) education (HR 1.29), Parkinson’s disease (HR 1.73), head injury (HR 1.37), peripheral vascular disease (HR 1.31), stroke (HR 1.29), depression (HR 1.18), alcohol intake (HR1.17), anaemia (HR 1.14), hypertension (HR 1.08), among other common comorbidities in HF (Figure 1A). Notably, a significant interaction (p < 0.001) between age and sex on dementia incidence was observed, such that women in all age groups were observed to have higher sHR compared to men (Figure 1B). After accounting for competing risk, dementia was not associated with adjusted hazard of all-cause mortality. Female sex, lower socioeconomic status, increasing age and common comorbidities were associated with higher hazards of incident dementia. Abstract Figure 1A and Figure 1B

Peak atrial longitudinal strain as an independent predictor of composite endpoint in patients received aortic valve replacement for severe aortic stenosis: a prospective cohort study

Abstract: Severe aortic stenosis (AS) is the primary valvular heart disease, treatable only by aortic valve replacement (AVR). The prognostic value of pre-operative left atrial (LA) function on post-AVR clinical outcomes is uncertain. The study aims to evaluate the prognostic value of pre-operative peak atrial longitudinal strain (PALS) as a surrogate of LA function on post AVR all-cause mortality and heart failure hospitalisation. Patients aged 18 years old or above with severe AS were recruited and assessed using speckle-tracking echocardiography pre-operatively. Severe AS was defined according to 2014 AHA/ACC Guideline for the Management of Patients with Valvular Heart Disease. PALS was measured. Based on the median value of PALS, patients were stratified into PALS <15.94% and PALS >15.94%. Patients with underlying pre-operative atrial fibrillation, other moderate to severe valvular heart diseases and cancers were excluded. Patients were followed up until death, heart failure hospitalisation or end of the study. The primary outcome is a composite endpoint of all-cause mortality and heart failure hospitalisation. The association of PALS with the composite endpoint was evaluated by Cox Proportional Hazards analysis. A total of 128 patients (mean age 65.32±9.42 years, 56.3% male) were prospectively analyzed. Patients were followed up for a mean period of 3.9±2.4 years. A total of 65 of 128 patients (50.8%) belonged to PALS<15.94%. During the study period, 23 patients developed the adverse events. A lower pre-operative PALS, both as a continuous or a categorical variable, were associated with a higher unadjusted risk of adverse events (Continuous; HR, 0.93; 95% CI 0.88–0.98; p=0.011; PALS <15.94%; HR, 4.94; 95% CI, 1.67–14.58; p=0.004). The study demonstrated a lower pre-operative PALS is associated with all-cause mortality and heart failure admission in patients with severe AS undergoing AVR. Evaluation of LA function by assessing speckle tracking derived PALS may aid in prognostication for patients undergoing AVR. Type of funding sources: None.

Comparison of risk of hyperkalemia between SGLT2 inhibitors and DPP4-inhibitors in patients with type 2 diabetes

Abstract: Type of funding sources: None. Hyperkalemia is a common complication and increases the risk of cardiac arrhythmias and mortality in patients with type 2 diabetes (T2DM), especially in those with diabetic nephropathy. We investigated the risk of hyperkalemia in patients initiated on SGLT2 inhibitors versus DPP-4 inhibitors among patients with T2DM. This study included patients with T2DM who initiated on SGLT2 inhibitors or DPP-4 inhibitors between January 01, 2015 and December 31, 2019 from a territory-wide clinical registry in Hong Kong (Clinical Data Analysis and Reporting System [CDARS]). A multivariable cox proportional hazards analysis, adjusting for key confounders, was used to compare the risk of central laboratory-determined hyperkalemia (serum potassium ≥6.0mmol/L) and hypokalemia (serum potassium <3.5mmol/L), respectively, between SGLT2 inhibitors and DPP-4 inhibitors. 10193 new users of SGLT2 inhibitors were matched to 17305 new users of DPP-4 inhibitors. During the 2-year follow-up, there were 104 hyperkalemia events (incident rate [IR] = 5.17 per 1000 person-years) among SGLT2 inhibitors and 306 events (IR = 9.09 per 1000 person-years) among DPP-4 inhibitors, of which SGLT2 inhibitors were associated with a lower risk of incident hyperkalemia (Adjusted HR: 0.66 [95%CI 0.53-0.83], p<0.001), compared to DPP-4 inhibitors. The incident hypokalemia was similar between SGLT2 inhibitors and DPP-4 inhibitors (Adjusted HR: 0.91 [95%CI 0.81-1.03], P=0.13). SGLT2 inhibitors reduced incident hyperkalemia, but without increasing incident hypokalemia compared to DPP-4 inhibitors.

Role of prenatal cardiovascular magnetic resonance imaging in determining pregnancy risk in repaired Tetralogy of Fallot patients

Abstract: Type of funding sources: None. Addressing pregnancy risks poses as a new challenge among women with repaired tetralogy of Fallot (TOF). The high-output state of pregnancy may predispose women to late complications of repaired TOF. However, guidelines regarding antenatal and or perinatal cardiovascular assessment has not been outlined. Noninvasive modalities such as cardiac magnetic resonance imaging (CMR) that do not require the utilization of ionizing radiation are feasible methods of assessment. Risk stratification of baseline CMR parameters has been sparsely investigated among repaired TOF cohorts. This study aims to identify baseline CMR parameters that may predict adverse outcomes of pregnancy among women with repaired TOF. Sixty-five successful pregnancies were recorded from a cohort of 105 pregnant episodes. Patients with CMR studies performed within 5 years prior to delivery events were included. Adverse clinical outcomes of interest included arrhythmia, heart failure admissions, gestational hypertension, pre-eclampsia toxemia and all-cause mortality. Baseline CMR parameters regarding ventricular mechanics such as left and right ventricular end-diastolic and end-systolic volumes, left and right ventricular ejection fractions and pulmonary regurgitant fraction were measured, and their association with adverse clinical outcomes were evaluated using an independent-samples t-test. Within all 65 successful pregnancies, 26 baseline CMR images were obtained and included in this study. The mean maternal age was 29.7 ± 6.7 years old with a mean birth weight of 2.91 ± 0.49 kg at a mean gestational age of 38.4 ± 2.0 weeks. There was a total of 12 patients with adverse clinical outcome: 5 patients with heart failure hospitalizations, 4 patients with gestational hypertension, 4 patients with arrhythmia and 2 patients with pre-eclampsia toxemia. Assessment of baseline CMR parameters of ventricular mechanics revealed that increased left ventricular end-diastolic volumes (LVEDV) (144.7mL/m2 ± 4.9; P = 0.011), left ventricular end-systolic volumes (LVESV) (65.1mL/m2 ± 9.4; P = 0.004), right ventricular end-diastolic volumes (RVEDV) (235.2 ± 29.0; P = 0.021) and right ventricular end-systolic volumes (RVESV) (122.2mL/m2 ± 38.1; P = 0.033) were associated with an increased incidence of arrhythmia during pregnancy. Nonetheless, there was no significant association between baseline ventricular mechanics with heart failure, gestational hypertension, and pre-eclampsia toxemia. Baseline assessment of cardiac magnetic resonance imaging among pregnant TOF women revealed that larger left and right ventricular volumes were associated with arrythmia development during pregnancy. Further studies with larger cohort sizes evaluating the role of antenatal and perinatal cardiovascular imaging assessment using CMR in predicting the risks of cardiovascular complications during pregnancy are warranted.

Diagnostic Criteria for Temporomandibular Disorders (DC/TMD) for Clinical and Research Applications: recommendations of the International RDC/TMD Consortium Network* and Orofacial Pain Special Interest Group†

Abstract: Temporomandibular disorders (TMD) are a significant public health problem affecting approximately 5% to 12% of the population.1 TMD is the second most common musculoskeletal condition (after chronic low back pain) resulting in pain and disability.1 Pain-related TMD can impact the individual's daily activities, psychosocial functioning, and quality of life. Overall, the annual TMD management cost in the USA, not including imaging, has doubled in the last decade to $4 billion.1 Patients often seek consultation with dentists for their TMD, especially for pain-related TMD. Diagnostic criteria for TMD with simple, clear, reliable, and valid operational definitions for the history, examination, and imaging procedures are needed to render physical diagnoses in both clinical and research settings. In addition, biobehavioral assessment of pain-related behavior and psychosocial functioning—an essential part of the diagnostic process—is required and provides the minimal information whereby one can determine whether the patient's pain disorder, especially when chronic, warrants further multidisciplinary assessment. Taken together, a new dual-axis Diagnostic Criteria for TMD (DC/TMD) will provide evidence-based criteria for the clinician to use when assessing patients, and will facilitate communication regarding consultations, referrals, and prognosis.2 The research community benefits from the ability to use well-defined and clinically relevant characteristics associated with the phenotype in order to facilitate more generalizable research. When clinicians and researchers use the same criteria, taxonomy, and nomenclature, then clinical questions and experience can be more easily transferred into relevant research questions, and research findings are more accessible to clinicians to better diagnose and manage their patients. The Research Diagnostic Criteria for Temporomandibular Disorders (RDC/TMD) have been the most widely employed diagnostic protocol for TMD research since its publication in 1992.3 This classification system was based on the biopsychosocial model of pain4 that included an Axis I physical assessment, using reliable and well-operationalized diagnostic criteria, and an Axis II assessment of psychosocial status and pain-related disability. The intent was to simultaneously provide a physical diagnosis and identify other relevant characteristics of the patient that could influence the expression and thus management of their TMD. Indeed, the longer the pain persists, the greater the potential for emergence and amplification of cognitive, psychosocial, and behavioral risk factors, with resultant enhanced pain sensitivity, greater likelihood of additional pain persistence, and reduced probability of success from standard treatments.5 The RDC/TMD (1992) was intended to be only a first step toward improved TMD classification, and the authors stated the need for future investigation of the accuracy of the Axis I diagnostic algorithms in terms of reliability and criterion validity—the latter involving the use of credible reference standard diagnoses. Also recommended was further assessment of the clinical utility of the Axis II instruments. The original RDC/TMD Axis I physical diagnoses have content validity based on the critical review by experts of the published diagnostic approach in use at that time and were tested using population-based epidemiologic data.6 Subsequently, a multicenter study showed that, for the most common TMD, the original RDC/TMD diagnoses exhibited sufficient reliability for clinical use.7 While the validity of the individual RDC/TMD diagnoses has been extensively investigated, assessment of the criterion validity for the complete spectrum of RDC/TMD diagnoses had been absent until recently.8 For the original RDC/TMD Axis II instruments, good evidence for their reliability and validity for measuring psychosocial status and pain-related disability already existed when the classification system was published.9–13 Subsequently, a variety of studies have demonstrated the significance and utility of the original RDC/TMD biobehavioral measures in such areas as predicting outcomes of clinical trials, escalation from acute to chronic pain, and experimental laboratory settings.14–20 Other studies have shown that the original RDC/TMD biobehavioral measures are incomplete in terms of prediction of disease course.21–23 The overall utility of the biobehavioral measures in routine clinical settings has, however, yet to be demonstrated, in part because most studies have to date focused on Axis I diagnoses rather than Axis II biobehavioral factors.24 The aims of this article are to present the evidence-based new Axis I and Axis II DC/TMD to be used in both clinical and research settings, as well as present the processes related to their development.

A classification of chronic pain for ICD-11.

Abstract: Chronic pain has been recognized as pain that persists past normal healing time5 and hence lacks the acute warning function of physiological nociception.35 Usually pain is regarded as chronic when it lasts or recurs for more than 3 to 6 months.29 Chronic pain is a frequent condition, affecting an estimated 20% of people worldwide6,13,14,18 and accounting for 15% to 20% of physician visits.25,28 Chronic pain should receive greater attention as a global health priority because adequate pain treatment is a human right, and it is the duty of any health care system to provide it.4,13 The current version of the International Classification of Diseases (ICD) of the World Health Organization (WHO) includes some diagnostic codes for chronic pain conditions, but these diagnoses do not reflect the actual epidemiology of chronic pain, nor are they categorized in a systematic manner. The ICD is the preeminent tool for coding diagnoses and documenting investigations or therapeutic measures within the health care systems of many countries. In addition, ICD codes are commonly used to report target diseases and comorbidities of participants in clinical research. Consequently, the current lack of adequate coding in the ICD makes the acquisition of accurate epidemiological data related to chronic pain difficult, prevents adequate billing for health care expenses related to pain treatment, and hinders the development and implementation of new therapies.10,11,16,23,27,31,37 Responding to these shortcomings, the International Association for the Study of Pain (IASP) contacted the WHO and established a Task Force for the Classification of Chronic Pain. The IASP Task Force, which comprises pain experts from across the globe,19 has developed a new and pragmatic classification of chronic pain for the upcoming 11th revision of the ICD. The goal is to create a classification system that is applicable in primary care and in clinical settings for specialized pain management. A major challenge in this process was finding a rational principle of classification that suits the different types of chronic pain and fits into the general ICD-11 framework. Pain categories are variably defined based on the perceived location (headache), etiology (cancer pain), or the primarily affected anatomical system (neuropathic pain). Some diagnoses of pain defy these classification principles (fibromyalgia). This problem is not unique to the classification of pain, but exists throughout the ICD. The IASP Task Force decided to give first priority to pain etiology, followed by underlying pathophysiological mechanisms, and finally the body site. Developing this multilayered classification was greatly facilitated by a novel principle of assigning diagnostic codes in ICD-11, termed “multiple parenting.” Multiple parenting allows the same diagnosis to be subsumed under more than 1 category (for a glossary of ICD terms refer to Table ​Table1).1). Each diagnosis retains 1 category as primary parent, but is cross-referenced to other categories that function as secondary parents. Table 1 Glossary of ICD-11 terms. The new ICD category for “Chronic Pain” comprises the most common clinically relevant disorders. These disorders were divided into 7 groups (Fig. ​(Fig.1):1): (1) chronic primary pain, (2) chronic cancer pain, (3) chronic posttraumatic and postsurgical pain, (4) chronic neuropathic pain, (5) chronic headache and orofacial pain, (6) chronic visceral pain, and (7) chronic musculoskeletal pain. Experts assigned to each group are responsible for the definition of diagnostic criteria and the selection of the diagnoses to be included under these subcategories of chronic pain. Thanks to Bedirhan Ustun and Robert Jakob of the WHO, these pain diagnoses are now integrated in the beta version of ICD-11 (http://id.who.int/icd/entity/1581976053). The Task Force is generating content models for single entities to describe their clinical characteristics. After peer review overseen by the WHO Steering Committee,39 the classification of chronic pain will be voted into action by the World Health Assembly in 2017. Figure 1 Organizational chart of Task Force, IASP, and WHO interactions. The IASP Task Force was created by the IASP council and its scope defined in direct consultation of the chairs (R.D.T. and W.R.) with WHO representatives in 2012. The Task Force reports to ... 2. Classification of chronic pain Chronic pain was defined as persistent or recurrent pain lasting longer than 3 months. This definition according to pain duration has the advantage that it is clear and operationalized. Optional specifiers for each diagnosis record evidence of psychosocial factors and the severity of the pain. Pain severity can be graded based on pain intensity, pain-related distress, and functional impairment. 2.1. Chronic primary pain Chronic primary pain is pain in 1 or more anatomic regions that persists or recurs for longer than 3 months and is associated with significant emotional distress or significant functional disability (interference with activities of daily life and participation in social roles) and that cannot be better explained by another chronic pain condition. This is a new phenomenological definition, created because the etiology is unknown for many forms of chronic pain. Common conditions such as, eg, back pain that is neither identified as musculoskeletal or neuropathic pain, chronic widespread pain, fibromyalgia, and irritable bowel syndrome will be found in this section and biological findings contributing to the pain problem may or may not be present. The term “primary pain” was chosen in close liaison with the ICD-11 revision committee, who felt this was the most widely acceptable term, in particular, from a nonspecialist perspective.

Diagnostic Criteria for Temporomandibular Disorders (DC/TMD) for Clinical and Research Applications: Recommendations of the International RDC/TMD Consortium Network* and Orofacial Pain

Abstract: Aims: The original Research Diagnostic Criteria for Temporomandibular Disorders (RDC/TMD) Axis I diagnostic algorithms have been demonstrated to be reliable. However, the Validation Project determined that the RDC/TMD Axis I validity was below the target sensitivity of ≥ 0.70 and specificity of ≥ 0.95. Consequently, these empirical results supported the development of revised RDC/TMD Axis I diagnostic algorithms that were subsequently demonstrated to be valid for the most common pain-related TMD and for one temporomandibular joint (TMJ) intra-articular disorder. The original RDC/TMD Axis II instruments were shown to be both reliable and valid. Working from these findings and revisions, two international consensus workshops were convened, from which recommendations were obtained for the finalization of new Axis I diagnostic algorithms and new Axis II instruments. Methods: Through a series of workshops and symposia, a panel of clinical and basic science pain experts modified the revised RDC/TMD Axis I algorithms by using comprehensive searches of published TMD diagnostic literature followed by review and consensus via a formal structured process. The panel’s recommendations for further revision of the Axis I diagnostic algorithms were assessed for validity by using the Validation Project’s data set, and for reliability by using newly collected data from the ongoing TMJ Impact Project—the follow-up study to the Validation Project. New Axis II instruments were identified through a comprehensive search of the literature providing valid instruments that, relative to the RDC/TMD, are shorter in length, are available in the public domain, and currently are being used in medical settings. Results: The newly recommended Diagnostic Criteria for TMD (DC/TMD) Axis I protocol includes both a valid screener for detecting any pain-related TMD as well as valid diagnostic criteria for differentiating the most common pain-related TMD (sensitivity ≥ 0.86, specificity ≥ 0.98) and for one intra-articular disorder (sensitivity of 0.80 and specificity of 0.97). Diagnostic criteria for other common intra-articular disorders lack adequate validity for clinical diagnoses but can be used for screening purposes. Inter-examiner reliability for the clinical assessment associated with the validated DC/TMD criteria for pain-related TMD is excellent (kappa ≥ 0.85). Finally, a comprehensive classification system that includes both the common and less common TMD is also presented. The Axis II protocol retains selected original RDC/TMD screening instruments augmented with new instruments to assess jaw function as well as behavioral and additional psychosocial factors. The Axis II protocol is divided into screening and comprehensive selfreport instrument sets. The screening instruments' 41 questions assess pain intensity, pain-related disability, psychological distress, jaw functional limitations, and parafunctional behaviors, and a pain drawing is used to assess locations of pain. The comprehensive instruments, composed of 81 questions, assess in further detail jaw functional limitations and psychological distress as well as additional constructs of anxiety and presence of comorbid pain conditions. Conclusion: The recommended evidence-based new DC/TMD protocol is appropriate for use in both clinical and research settings. More comprehensive instruments augment short and simple screening instruments for Axis I and Axis II. These validated instruments allow for identification of patients with a range of simple to complex TMD presentations. J Oral Facial Pain Headache 2014;28:6–27. doi: 10.11607/jop.1151

Pathophysiology of Migraine: A Disorder of Sensory Processing

Abstract: Plaguing humans for more than two millennia, manifest on every continent studied, and with more than one billion patients having an attack in any year, migraine stands as the sixth most common cause of disability on the planet. The pathophysiology of migraine has emerged from a historical consideration of the "humors" through mid-20th century distraction of the now defunct Vascular Theory to a clear place as a neurological disorder. It could be said there are three questions: why, how, and when? Why: migraine is largely accepted to be an inherited tendency for the brain to lose control of its inputs. How: the now classical trigeminal durovascular afferent pathway has been explored in laboratory and clinic; interrogated with immunohistochemistry to functional brain imaging to offer a roadmap of the attack. When: migraine attacks emerge due to a disorder of brain sensory processing that itself likely cycles, influenced by genetics and the environment. In the first, premonitory, phase that precedes headache, brain stem and diencephalic systems modulating afferent signals, light-photophobia or sound-phonophobia, begin to dysfunction and eventually to evolve to the pain phase and with time the resolution or postdromal phase. Understanding the biology of migraine through careful bench-based research has led to major classes of therapeutics being identified: triptans, serotonin 5-HT1B/1D receptor agonists; gepants, calcitonin gene-related peptide (CGRP) receptor antagonists; ditans, 5-HT1F receptor agonists, CGRP mechanisms monoclonal antibodies; and glurants, mGlu5 modulators; with the promise of more to come. Investment in understanding migraine has been very successful and leaves us at a new dawn, able to transform its impact on a global scale, as well as understand fundamental aspects of human biology.

Global, regional, and national burden of migraine and tension-type headache, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016

Abstract: Summary Background Through the Global Burden of Diseases, Injuries, and Risk Factors (GBD) studies, headache has emerged as a major global public health concern. We aimed to use data from the GBD 2016 study to provide new estimates for prevalence and years of life lived with disability (YLDs) for migraine and tension-type headache and to present the methods and results in an accessible way for clinicians and researchers of headache disorders. Methods Data were derived from population-based cross-sectional surveys on migraine and tension-type headache. Prevalence for each sex and 5-year age group interval (ie, age 5 years to ≥95 years) at different time points from 1990 and 2016 in all countries and GBD regions were estimated using a Bayesian meta-regression model. Disease burden measured in YLDs was calculated from prevalence and average time spent with headache multiplied by disability weights (a measure of the relative severity of the disabling consequence of a disease). The burden stemming from medication overuse headache, which was included in earlier iterations of GBD as a separate cause, was subsumed as a sequela of either migraine or tension-type headache. Because no deaths were assigned to headaches as the underlying cause, YLDs equate to disability-adjusted life-years (DALYs). We also analysed results on the basis of the Socio-demographic Index (SDI), a compound measure of income per capita, education, and fertility. Findings Almost three billion individuals were estimated to have a migraine or tension-type headache in 2016: 1·89 billion (95% uncertainty interval [UI] 1·71–2·10) with tension-type headache and 1·04 billion (95% UI 1·00–1·09) with migraine. However, because migraine had a much higher disability weight than tension-type headache, migraine caused 45·1 million (95% UI 29·0–62·8) and tension-type headache only 7·2 million (95% UI 4·6–10·5) YLDs globally in 2016. The headaches were most burdensome in women between ages 15 and 49 years, with migraine causing 20·3 million (95% UI 12·9–28·5) and tension-type headache 2·9 million (95% UI 1·8–4·2) YLDs in 2016, which was 11·2% of all YLDs in this age group and sex. Age-standardised DALYs for each headache type showed a small increase as SDI increased. Interpretation Although current estimates are based on limited data, our study shows that headache disorders, and migraine in particular, are important causes of disability worldwide, and deserve greater attention in health policy debates and research resource allocation. Future iterations of this study, based on sources from additional countries and with less methodological heterogeneity, should help to provide stronger evidence of the need for action. Funding Bill & Melinda Gates Foundation.