Surgery for children in low-income countries affected by humanitarian emergencies from 2008 to 2014: The Médecins Sans Frontières Operations Centre Brussels experience.

All procedures performed in an operating theater managed by MSF-OCB from July 2008 through December 2014 were recorded using a standardized Patient Surgical Record (PSR). The PSR was developed by MSF for operational research needs. Therefore, patient identifiers for longitudinal evaluation and specific clinical data, such as imaging, laboratory values and long-term outcomes were not recorded. The PSR variables included age, gender, American Society of Anesthesia (ASA) physical status score, type of anesthesia utilized, operative indication, procedure(s) performed, procedure order and urgency, perioperative death (i.e. death from initiation of anesthesia care to discharge from the recovery ward) and project site. The PSR was transcribed monthly into a database (Excel; Microsoft, Redmond, WA, USA) and transmitted to MSF-OCB headquarters in Brussels, Belgium. At headquarters, the Surgical, Anesthesia, Gynecology and Emergency Medicine (SAGE) Unit reviewed all data for completeness and accuracy. Discrepancies, missing data or questions were immediately corrected after reconciliation with program personnel. The reason for MSF-OCB assistance for each project was characterized as maternity, natural disaster, conflict, or hospital support. Projects primarily dedicated to the care of women during childbirth and in the peri-partum setting were considered maternity. Non-maternity projects that did not care for those injured as a result of widespread conflict or after natural disaster were considered to be hospital support. Children were defined as all individuals under the age of 18 years. Age groups were defined using the Centers for Disease Control and Prevention epidemiologic health-related age groupings (i.e. <1 year, 1–4 years, 5–9 years, 10–14 years and 15–24 years) [ 25 ]. The 15–24 year old age group was truncated at 18 years of age. Types of anesthesia included general anesthesia with an airway (i.e. endotracheal intubation, laryngeal mask airway), general anesthesia without an airway (i.e. ketamine-based anesthesia), regional or local anesthesia (e.g. spinal, epidural, peripheral nerve blocks, local infiltration), and ‘combined or other.’ Procedures were defined as emergent, urgent, or elective; and were classified as first operation, planned return operation, or unplanned return operation. MSF classification groups and sub-groups for indication and procedure type have been detailed elsewhere [ 20 ]. Briefly, indication for operation was grouped into one of three mutually exclusive categories: trauma, non-trauma, and obstetric/gynecologic (Ob/Gyn). Trauma indications included injuries associated with conflict (e.g., those resulting from mines, bombs, gun shots, assault, gender-based violence, torture), and injuries not directly associated with conflict (e.g., those from road traffic collisions, falls, burns). Non-trauma indications included abscess, typhoid perforations, intestinal obstruction, appendicitis, tumor, and others. Ob/Gyn indications included maternal fetal conditions, referred to in this manuscript as maternal/fetal to avoid confusion between indication for operation and procedure description. Procedures were grouped into six mutually exclusive categories: minor surgery, wound surgery, visceral surgery, orthopedics, obstetrics/gynecology/urology (Ob/Gyn/Uro), and specialized surgery. Minor surgery, wound surgery, and visceral surgery were considered “general surgical” procedures. Procedure categories have also been detailed previously [ 20 ]. Examples of minor surgery included simple wound treatment, abscess drainage, circumcision and drain placement. Wound surgery included extensive debridement, digital amputation, burn dressings, and foreign body removal. Visceral surgery included operations such as appendectomy, exploratory laparotomy and repair of solid or hollow viscus injury. Orthopedic surgery included treatment of fractures, amputations, joint surgery, and repair of tendon or nerve injury. Ob/Gyn/Uro surgery included cesarean sections, genital fistula procedures, tubal ligation, and interventions for renal and external genital pathology. Specialized surgery included procedures typically completed under the auspices of ophthalmology, neurosurgery, maxillofacial surgery and others [ 20 ]. Total and age-specific counts and/or relative proportions were presented for patient characteristics, indication for operation and operative details. Further analysis of surgery for infants (i.e., children <1 year old) and surgeries for trauma was presented to highlight important aspects of caring for these vulnerable populations in austere settings and during conflict or after disaster. Chi 2 test for independence (with degrees of freedom equal to k R − 1 row categories times k C − 1 column categories) was used to evaluate differences between categorical variables. One- and two-sample tests of proportions with two-sided p-value set at 0.05 were completed for comparison of binomial proportions. Bivariate logistic regression was performed to evaluate predictors of perioperative death. A multivariable logisticmodelwas built based on bivariate statistical significance at the p = 0.05 level and an a priori hypothesis of the following clinically relevant confounders: age, sex, ASA score, and indication for operation (trauma, non-trauma, maternal/fetal). All analyses were completed with Stata v12 (College Station, TX, USA). To compare the relative proportion of procedures performed between age groups, ranks are provided. Ranks were shaded to more easily appreciate the change in rank between age groups. For all rank lists, dark gray is the most common procedure performed. Procedures performed less commonly were shaded progressively lighter shades of gray. White represents the least common procedure performed. Retrospective description of de-identified, routinely collected data satisfied MSF Ethical Review Board exemption criteria. Johns Hopkins Bloomberg School of Public Health Institutional Review Board approved secondary analyses.

Between 2008 and 2014, MSF-OCB provided surgical assistance at 45 projects in 19 countries ( Fig. 1 ). In total, 109,828 operations were performed during the study period; 24,576 were for children (22%). Twenty-one projects were conflict missions (47% of projects). Among conflict projects, nearly one-third of the operations performed were for children (13,969; 29%). Twelve projects provided hospital support (27%), six provided assistance after natural disaster (13%), and six were maternity missions (13%); these projects performed 21%, 19% and 8% of operations for children, respectively. Two-thirds of children were male (15,327; 62%) and the mean age was 8.9 years (SD 5.2; Table 1 ). When stratified by age group, demographic and operative characteristics varied significantly (p < 0.001 for sex, ASA category, anesthesia type, surgical indication, operative urgency, and perioperative death). Notably, among 15–17 year olds, girls made up a greater proportion of surgeries than boys (55% vs. 45%, p < 0.001). In this age group, maternal/fetal conditions accounted for one-third of operative indications (n = 1723; 34% of operations for 15–17 year olds). Trauma was the most common indication for surgery outside of infancy. Trauma operations were nearly twice as common compared to non-trauma operations in children over five (5–9 years, 65% vs. 35%; 10–14 years, 65% vs. 33%; 15–17 years 41% vs. 25%; p < 0.001), while only slightly more common in children 1–4 years of age (57% vs. 42%; p < 0.001). Seventy-one percent of operations for infants were for non-trauma indications. In infants, skin and soft tissue infection accounted for almost half of all operative indications (n = 504; 46%). Other common causes included intestinal obstruction, hernia, and circumcision, which together accounted for 17% of operative indications among infants. Congenital anomaly was the indication for 6% of operations for infants (64 of 1096 operations) and only 1% of operations for children 1–4 years of age (63 of 5337 operations). There were 26,284 procedures performed for the 24,576 children during the study period ( Table 2 ). Of these, 5539 (21%) were minor (e.g. simple wound closure, incision and drainage, circumcision). Extensive wound debridement and procedures for burn care were also common (16% and 11% of procedures, respectively). The majority of procedures were general surgical (19,582; 75%); however, 17%were orthopedic, 8% were Ob/Gyn/Urologic, and 1% were more specialized. The most salient difference between procedures by age group was the greater relative proportion of burn care procedures in younger children (19% <1 year, 23% 1–4 years, 12% 5–9 years, 5% 10–14 years, and 3% 15–17 years; p < 0.001; Table 2 ; expanded list provided in Appendix A ). Since trauma was the most common indication for surgery (13,984 operations; 57%), this was further explored. Predictably, trauma accounted for the majority of operations for children at projects providing assistance after natural disaster or during conflict (79% and 66% of all operations, respectively). However, 38% of operations at hospital support projects and 3% of operations at maternity projects were for trauma. The proportion of operations for trauma increased among all project-types during the study period (27% to 73% of all operations performed in 2008 and 2014, respectively; p < 0.001) ( Fig. 2 ; detailed list of procedural indications by project-type provided in Appendix B ). The increase was driven by unintentional injuries (21% of all pediatric operations in 2008 to 63% in 2014; p < 0.001) more than conflict injuries (6% in 2008 to 10% in 2014; p < 0.001), despite the change in relative proportion reaching statistical significance in both categories. Injury mechanism varied by age group ( Table 4 ). Most trauma operations were for mechanisms not directly related to conflict (11,664/13,984 trauma operations; 83%). Road traffic collisions accounted for 22%, burns 24%, and other injuries including falls 37%. Violence and conflict-related mechanisms accounted for 17% of trauma operations (gunshot wounds 8%; bomb blasts 5%; stab wounds 2%; and others 1%). Violence contributed to a greater proportion of trauma operations with each increase in age group category: 7% of infants, 8% of children 1–4 years, 10% 5–9 years, 22% 10–14 years, and 31% 15–17 years ( p < 0.001 for trend). The most frequently performed procedures after injury included the management of complex wounds (e.g., extensive debridement, burn care, skin/muscle graft; 6764 procedures; 44% of all trauma procedures), fractures (3319; 21%), and minor wounds (e.g., minor debridement, simple wound closure; 3549; 23%). There were 42 pediatric perioperative deaths during the study period (i.e. death from start of anesthesia care to discharge from the recovery room; perioperative death rate 0.17%; detailed demographic and procedural details provided in Appendix C ). Perioperative death was significantly higher in infants (0.82%; 9/1096 infants) compared to all other age groups (1–4 years, 0.11%; 5–9 years, 0.12%; 10–14 years, 0.16%; 15–17 years, 0.16%; p < 0.001; Table 1 ). In bivariate analysis comparing children who died compared to children who survived, age, urgency of operation, ASA score, type of anesthesia, and reason for humanitarian assistance were important predictors of perioperative death ( Table 3 ). Gender and indication for operation (i.e., trauma, non-trauma, maternal/fetal) were not associated with death. The final multivariable logistic model, which controlled for age, sex, urgency of operation, ASA, type of anesthesia, indication for operation and reason for humanitarian assistance, showed high discrimination (c-statistic 0.9343) and modest goodness of fit (McFadden’s pseudo-R 2 = 0.36). Age less than one year, ASA score ≥3, emergency operation, general anesthesia with an airway, and conflict project were independently associated with death ( Table 3 ). No perioperative deaths occurred during elective operations or at maternity projects. Each of the nine infants who died was treated during conflict, however only three were related to trauma (i.e., two gunshot wounds, one severe burn). Three additional infant deaths were from congenital disease (e.g. giant omphalocele, gastroschisis, undefined), two from intussusception, and one from peritonitis of unknown etiology.

This study aimed to describe surgical care for children at MSF-OCB projects from 2008 to 2014, with a focus on age-specific procedure distributions and trauma surgery. Surgery for children during humanitarian crises was very common, particularly during conflict. Traumatic injuries were most common across all project-types, but non-trauma and maternal/fetal pathology were also prevalent. The most common procedures require a variety of operative skills, from debridements and bowel resections to cesareans and fracture fixation. Additionally, this study demonstrated that infants, as well as children undergoing general anesthesia with an airway, had greater odds of death than other children undergoing surgery. These findings are useful for humanitarian surgical assistance programs planning to strategically allocate resources and clinicians embarking on humanitarian missions. This study also contributes to the evidence base that supports the need to advocate for the protection of children during war. Several descriptions of the operative epidemiology of children treated by humanitarian projects have been published [ 16 , 26 – 32 ]. However, most of these refer to specialty- and/or country-specific projects in cardiac surgery, neurosurgery, otolaryngology, and ophthalmology during non-disaster, non-conflict time periods [ 16 , 26 – 32 ]. Recently, descriptions of surgical care in post-earthquake Haiti and during Operation Iraqi Freedom and Operation Enduring Freedom have also been published [ 32 , 33 , 36 ]. However, large, multi-country descriptions of pediatric surgical care from humanitarian surgical assistance projects during crisis have not been published. Continuing Promise, a surgical assistance project of the United States Navy, deployed to seven peace-time countries in Central and South America, and treated 340 children between April and July 2009 [ 31 ]. The most frequently performed procedures were inguinal and umbilical hernias (23% and 14%, respectively), followed by circumcision (9%), mass excision (8%), and hydrocele/varicocele/orchiopexy (5%) [ 31 ]. Surgical sub-specialists performed 36% of cases [ 31 ]. In disaster settings, orthopedic injuries and wounds account for a greater proportion of pediatric surgical cases than before a disaster. For example, the Israeli Defence Force Medical Corps dispatched a field hospital after the 2010 Haiti earthquake. The team treated 318 pediatric patients in 10 days; 57 (21%) required surgery, most commonly performing wound debridement (20%), external fixation (16%) and amputation (14%) [ 37 ]. The United States Naval Ship Comfort treated 237 pediatric surgical patients in Haiti after the earthquake; orthopedic procedures represented 71% of the cases [ 38 ]. Partners in Health/Zanmi Lasante-run hospitals in Haiti reported on procedures before and after the earthquake. About 21% of operations completed between July 2009 and July 2010were for children [ 33 ]. In pre-earthquake Haiti, congenital anomalies accounted for 34% of procedures, hernia 21%, and genitourinary and obstetric causes 14% each, while trauma and burns accounted for only 3% of pediatric operations [ 33 ]. In post-earthquake Haiti, trauma and burns accounted for 39% of pediatric operations [ 33 ]. These reports demonstrate that the scope of humanitarian surgical care during and after crisis differs significantly from that in usual LMIC settings. Our study moved beyond single-country descriptions, and explored in detail the breadth and depth of surgical procedures performed on children at a variety of project-types in countries around the world, all with different crises and struggles, making our findings more generalizable and informative for the physician and trainee interested in humanitarian work and global surgery as well as for foundation-based resources allocation. MSF-OCB projects do not focus on pediatric surgery specifically. Many of the procedures that may fall under a pediatric surgeon’s purview in a high-income country are completed by a general surgeon or a medical officer in LMICs and during humanitarian crisis. Therefore, outside of emergency procedures for congenital anomalies (e.g. neonatal intestinal obstruction), surgeon availability may drive the relative proportions of operative indications performed in our study population. If more pediatric surgeons participated in MSF projects, more congenital anomalies could be addressed, considerate of perioperative capacity. Ultimately, pediatric surgical care capacity building is required to reduce unmet surgical care needs in children [ 34 ]. This includes medical and nursing education, capital investment in infrastructure (e.g. essential surgical care equipment and supplies, incubators, pediatric-sized nondrug consumables), creation and maintenance of a reliable supply chain, and dependable financing mechanisms. In the meantime, dedicated pediatric surgical assistance programs are required to augment current demand in LMICs by providing patient care and on-the-job training to local staff. [ 35 ] A few particular findings in our study warrant additional comment: in particular, MSF-OCB projects cared for a significant number of burn injuries, particularly in younger ages. Children in LMICs are at high-risk of burn injury for a number of reasons; particularly, proximity to open cooking fires. Additionally, safe infrastructure and behavior that typically protect children from injury are often disrupted during conflict or after natural disaster [ 39 ]. These data corroborate findings from a recent review on burns in LMICs, wherein burn injury was more common among infants and toddlers from birth to 4 years of age, and then tapered off during adolescence [ 38 ]. Pediatric burn care demands significant resources from surgical projects compared to other conditions given the need for serial procedures (e.g., extensive debridements), large dressing changes under anesthesia, intensive care and long hospitalizations [ 40 ]. Therefore, anticipating this burden and ensuring that the knowledge, skills and resources necessary to adequately care for pediatric injury and burns are available to humanitarian surgical projects are important. Salient in our findings, about one in every ten children who underwent surgery at MSF-OCB projects was a victim of land mines, bombs, firearms, assault, stabbings, gender-based violence or torture. Humanitarian surgical teams are in a unique position to identify a pattern of intolerable acts affecting children given their exposure to these injuries and understanding of the contextual complexities. Subsequently, surgical teams have an opportunity to advocate for the protection of children and other vulnerable populations during conflict. Examples of successful healthcare-led recognition and systematic documentation of such acts that built an evidence-base for advocacy that, in turn, led to significant steps toward civilian protection include the Mine Ban Treaty, Convention on Cluster Munitions and the conference of the Ethiopian Society of Obstetricians and Gynecologists and Synergie des Femmes pour les Victimes des Violences Sexuelles [ 41 – 43 ]. These agreements or meetings provided a platform for the creation of international legislation, amendments to the rules of war and the establishment of specific injury management guidelines. Congenital operations are beyond the intended scope of MSF-OCB projects. However, the existing literature supports a role for surgical assistance programs to reduce this burden when not operating during crisis and when specialist care is available [ 4 , 19 , 33 , 44 ]. Higashi and colleagues described a significant burden of disease from common congenital anomalies in LMICs (e.g., cleft lip and palate, intestinal atresia, neural tube defects) [ 19 ], wherein 57% of the 21.6 million DALYs incurred by congenital anomalies may potentially be addressed by access to surgical care [ 19 ]. Gupta et al. found that 7% of all unmet surgical need in Nepal was due to congenital anomalies [ 45 ]. Groen et al. estimated that need to be 3% in Sierra Leone [ 46 ]. MSF-OCB refrains from performing surgery for congenital anomalies when possible for two reasons: first, there is an extraordinary demand for emergency procedures (e.g., injury, acute abdomen); and second, there is a lack of humanitarian staff with the experience or training necessary to care for complex anomalies (e.g., cleft lip, imperforate anus). However, humanitarian surgical assistance programs that do not operate during crisis may consider addressing the congenital disease burden and building pediatric surgical capacity as part of their directive. The perioperative death rate was low in MSF-OCB projects. However, perioperative death among infants was significantly higher compared to older age groups. Independent risk factors for perioperative death were identified: ASA score ≥3, emergency operation, general anesthesia with an airway, and conflict project. Most of these risk factors were not modifiable at the hospital-level. However, particular care should be taken when deciding to operate on an infant or determining that a child requires an airway. The perioperative capacity of many hospitals and humanitarian projects in LMICs is extremely limited [ 15 ]. As a result, teams often work with few physical resources and rely on task-sharing [ 47 , 48 ]. While the outcomes of projects that rely on task-sharing are similar to projects with a well-trained anesthetist [ 48 ], complex cases (e.g., surgery for infants, management of children in extremis) require greater resources, training and experience. Taking these into consideration when resourcing projects and training task-sharers is important. Though this study represents the largest description of surgeries for children during humanitarian crises to date, the data must be interpreted cautiously. The operational data used for this study do not include certain variables that would be helpful for more accurately estimating resource needs for future humanitarian projects (e.g., prehospital deaths, those that did not seek care but had a surgical condition, children with conditions too complex than resources could manage or in too great a number to be treated, specific-item consumption (e.g. gauze use for burn care, plaster use for fractures)). Procedural details and specifics related to operative indication(s) were limited to the coding structure of the PSR [ 20 ]. Similarly, detailed clinical data that could better guide quality improvement programs were not collected (e.g., time to theatre after diagnosis, post-operative pain assessments and longer-term outcome variables, such as in-hospital death or discharge disability scores). The importance of incorporating these variables into routine data collection by surgical projects cannot be overstated. However, doing so with limited number of staff and in projects that typically operate above capacity is complex and potentially detrimental to the quality or efficiency of care in the short run. Lastly, the perioperative death rate reported in this study was unexpectedly low and does not include the mortality related to surgical disease in the pediatric population more broadly. This was likely the result of two factors: i) many children died before reaching the hospital or were not operated on because they would have required care beyond the capabilities of the facility at that time; and ii) the indicator does not incorporate deaths after discharge from the recovery room. The death rate of children during crisis is certainly higher than what is reported here. Despite these limitations, these results allow reasonable conclusions to be drawn with regard to the large burden of pediatric surgical disease in LMICs during crisis and the need to incorporate pediatric-specific considerations into the planning, monitoring and evaluation processes of humanitarian surgical assistance programs. Further, by describing pediatric operative epidemiology across countries, projects and time, our findings are more generalizable than hospital- or country-specific descriptions previously published. Nearly a quarter of surgical procedures performed at MSF-OCB projects were for children, with injury being the most common pediatric operative indication. The case mix was diverse and required a broad skill-set. Among the many findings that require consideration in current and future humanitarian surgical projects are the greater odds of death among infants and children who required an airway as well as the considerable unmet surgical need for congenital anomalies. This study identified several ways in which the humanitarian surgical community can potentially reduce the death and disability of children living in LMICs during crisis: better allocate resources to hospitals and humanitarian surgical assistance projects appropriately, with particular consideration of pediatric-sized supplies; improve the recognition of high-risk children and training for non-pediatric surgeons and anesthetists caring for this age group; and advocate for the role of surgery in public health, for the prevention of injuries among children affected by the breakdown of safe society, and for the protection of vulnerable populations during conflict. By responding to these challenges, the death and disability of children living in LMICs affected by crisis might be reduced and a safer, healthier and more productive generation promoted.