Journal of Trauma and Acute Care Surgery:
WTA 2013 Plenary Paper
Gunshot wounds and blast injuries to the face are associated with significant morbidity and mortality: Results of an 11-year multi-institutional study of 720 patients
Shackford, Steven R. MD; Kahl, Jessica E. BA; Calvo, Richard Y. PhD(c); Kozar, Rosemary A. MD; Haugen, Christine E. BA; Kaups, Krista L. MD; Willey, Marybeth BA; Tibbs, Brian M. MD; Mutto, Susan M. RN; Rizzo, Anne G. MD; Lormel, Christy S. MPH; Shackford, Meghan C. BA; Burlew, Clay Cothren MD; Moore, Ernest E. MD; Cogbill, Thomas H. MD; Kallies, Kara J. BA; Haan, James M. MD; Ward, Jeanette BA
From the Scripps Mercy Hospital (S.R.S., J.E.K., R.Y.C., M.C.S.), San Diego; and University of California San Francisco/Fresno (K.L.K., M.W.), Fresno, California; University of Texas Houston and Memorial Hermann Hospital (R.A.K., C.E.H.), Houston, Texas; Mercy Hospital (BMT, SMM), Springfield, Missouri; Inova Fairfax Hospital (A.G.R., C.S.L.), Falls Church, Virginia; Denver General Health and University of Colorado (C.C.B., E.E.M.), Denver, Colorado; Gundersen Lutheran Medical Center (T.H.C., K.J.K.), La Crosse, Wisconsin; and Via Christi Hospital (JMH, JW), Wichita, Kansas.
Submitted: February 14, 2013, Revised: August 28, 2013, Accepted: September 15, 2013. Published online: January 6, 2014.
This study was presented at the 43rd Annual Meeting of the Western Trauma Association, March 3–8, 2013, in Snowmass, Colorado.
Address for reprints: Steven R. Shackford, MD, Trauma Service, Scripps Mercy Hospital, 4077 5th Ave, San Diego, CA 92103; email:
Gunshot wounds and blast injuries to the face (GSWBIFs) produce complex wounds requiring management by multiple surgical specialties. Previous work is limited to single institution reports with little information on processes of care or outcome. We sought to determine those factors associated with hospital complications and mortality.
We performed an 11-year multicenter retrospective cohort analysis of patients sustaining GSWBIF. The face, defined as the area anterior to the external auditory meatuses from the top of the forehead to the chin, was categorized into three zones: I, the chin to the base of the nose; II, the base of the nose to the eyebrows; III, above the brows. We analyzed the effect of multiple factors on outcome.
From January 1, 2000, to December 31, 2010, we treated 720 patients with GSWBIF (539 males, 75%), with a median age of 29 years. The wounding agent was handgun in 41%, explosive (shotgun and blast) in 20%, rifle in 6%, and unknown in 33%. Prehospital or resuscitative phase airway was required in 236 patients (33%). Definitive care was rendered by multiple specialties in 271 patients (38%). Overall, 185 patients died (26%), 146 (79%) within 48 hours. Of the 481 patients hospitalized greater than 48 hours, 184 had at least one complication (38%). Factors significantly associated with any of a total of 207 complications were total number of operations (p < 0.001), Revised Trauma Score (RTS, p < 0.001), and head Abbreviated Injury Scale (AIS) score (p < 0.05). Factors significantly associated with mortality were RTS (p < 0.001), head AIS score (p < 0.001), total number of operations (p < 0.001), and age (p < 0.05). An injury located in Zone III was independently associated with mortality (p < 0.001).
GSWBIFs have high mortality and are associated with significant morbidity. The multispecialty involvement required for definitive care necessitates triage to a trauma center and underscores the need for an organized approach and the development of effective guidelines.
LEVEL OF EVIDENCE
Therapeutic/care management, level III.
Gunshot wounds and blast injuries to the face (GSWBIFs) are devastating injuries that can be life threatening, with reported mortality rates as high as 15%.1 If patients survive, morbidity is high, with complication rates of up to 30%.2–5 Despite this relatively high rate of morbidity and mortality, there have been no comprehensive descriptive reviews of the processes of care involved with managing these complex injuries and only one recent report that addresses overall outcome.5 Furthermore, facial injuries are definitively managed by different specialties (e.g., plastic surgery, otolaryngology, oral and maxillofacial surgery), which introduces practice variation, complicating interpretation of the effect of management on outcome. Moreover, information exchange is impeded owing to publication in specialty-specific journals. These reports focus on procedural or technical issues unique to the specialty, rather than on the full spectrum of treatment necessary to manage these complex injuries. Finally, there is no consistent or standard definition of the anatomic boundaries of the face. Although many reports include the region above the orbits in their definition,6–9 others exclude it.1,10–13 Still, others fail to describe any anatomic boundaries of the face.2,14,15
For these reasons and because most centers encounter few such wounds in the course of a year, GSWBIFs are well suited to a retrospective multicenter review. Such a review of the management and outcomes would have application not only to civilian trauma centers but also to the military since injuries of the face and head are increasingly common in the theater of operation.16,17
We performed an 11-year multicenter retrospective descriptive review of the care and outcomes of all patients with GSWBIF. Our purpose was to identify and describe those factors that were associated with adverse outcome.
PATIENTS AND METHODS
Data were gathered from eight centers participating in the Western Trauma Association Multicenter Trials Committee. Seven centers were verified by the American College of Surgeons’ Committee on Trauma; six were Level I trauma centers, and one was Level II. One center was state verified.
After obtaining institutional review board approval at each center, we reviewed the records of all patients admitted between January 1, 2000, and December 31, 2010, with either a gunshot wound or a blast injury of the face. For study purposes, injuries resulting from a gunshot included those from a handgun, rifle, or shotgun. Patients who sustained injuries from a pellet gun, compressed air device (nail gun), or compressed air rifle were excluded. Injuries resulting from an explosion or an explosive device and coded as such in the discharge data set were considered to be caused by a blast. The records of patients admitted with the following E-codes were reviewed: 921–922.3, 922.5–923.9, 955–955.5, 965–968.5, 970, 971, 979.2, 979.4, 985, 991–991.9, 993–993.9, 998, and 998.9. To be inclusive, we also reviewed the records of all patients with complex open fractures of the face with the following DRG International Classification of Diseases—9th Rev. (ICD-9) codes: 800.5–800.9, 801.5–801.9, 802.1, 802.30–802.39, 802.5, 802.7, 803.5–803.9, 804.5–804.9, and 870.0–873.9. Shotgun wounds and blast injuries were combined for analysis into a single category, “blast-type injury.”
We defined the face as the area bounded laterally by the external auditory meatuses, superiorly by the hairline (or where the angle between the forehead and caput changes) and inferiorly by the chin as previously described by Gant and Epstein6 as well as Gussack and Jurkovich.7 Patients with associated gunshot wounds to other areas (i.e., wounds to the abdomen, chest, or extremities) were included but analyzed separately.
For the purpose of analysis, we divided the face into three zones (Fig. 1). Zone 1 was bounded by the chin inferiorly and the base of the nose superiorly. Zone 2 was bounded by the base of the nose inferiorly and the supraorbital ridge superiorly. Zone 3 was superior to the supraorbital ridge.
Data abstracted from the medical record included patient demographics, comorbidities, mechanism of injury, physiology at admission, blood and urine toxicology results, severity of facial and associated injuries, management, outcomes, and specialty certification of surgeons involved with care (Table 1). From the abstracted comorbidities, a Charlson score was calculated.18 Data were entered on a standardized form or directly on an Excel spreadsheet (Microsoft, Redmond, WA) and submitted to a central repository (Scripps Mercy Hospital, San Diego, CA). Data from each center were reviewed on two separate occasions by three of the authors (S.R.S., J.E.K., M.C.S.) before entry into a composite file. Queries regarding the data were addressed by e-mail and telephone conferences to assure that there was homogeneity in the data (e.g., procedures, associated injuries, missing values). If there was any concern about the validity of a data point, it was considered to be missing. As a result, data on antibiotic use were missing in 11% (4% among patients surviving >48 hours). In the remaining fields, missing values ranged from less than 1% to 4%. Missing data were treated as voids, not as zeroes, reducing the number of total entries in the particular field.
Two adverse outcomes were defined for this study, namely, mortality and incident complications. We further categorized incident complications into complications of injury and complications of care. Complications of injury included complications thought to be immediate and occurring before admission as a consequence of the injuring agent (e.g., blindness from a direct globe injury). Complications of care included complications occurring after admission or later during treatment as a result of either disease progression or a consequence of care (e.g., wound infection).
General descriptive statistics were obtained using Student’s t test and χ2 analyses. Binary logistic regression was used to identify covariates associated with each of the outcomes independently. For each outcome, logistic mixed models were fit using unstructured covariance matrices and with the hospital of origin as the random effect. These models evaluated the relationship between risk factors and the outcomes after accounting for the unobserved effects related to the hospital of origin. Data were analyzed with Stata/MP version 11.2 (StataCorp LP, College Station, TX). Statistical significance was attributed to a p < 0.05.
During the 11-year study period, the eight centers treated 720 patients with GSWBIF (Table 2). On average, individual centers treated between less than 1 and 20 GSWBIFs per year. Most of the patients were young males with isolated, moderately severe craniofacial injuries inflicted by a handgun. Of the 684 patients with toxicology data available, 108 (16%) had toxicology screens positive for cocaine, methamphetamine, lysergic acid diethyamide (LSD) cannabis, or heroin; 218 (31%) of 707 patients with data available had a positive blood alcohol level. Ninety-one patients (13%) had severe associated injuries in regions other than the neck or face. In addition to their facial injuries, 12 had combined injuries of the chest and abdomen, 40 had injuries to the chest only, 13 had injuries to the abdomen only, and 36 had injuries to the extremities only. Forty-eight patients had associated injuries in the neck.
Comorbidities were present in 98 patients (14%). However, the average Charlson score was less than 1. Seven patients (1%) had a psychiatric illness, and 70 (10%) had a mental disorder (e.g., depression, bipolar disorder) documented in their medical record.
During the prehospital and resuscitative phase, urgent airway control was necessary in 236 patients (33%). This was achieved by endotracheal intubation in 217 (92%), nasotracheal intubation in 6 (2.5%), cricothryroidotomy in 12 (5%), and tracheostomy in 1 (0.4%). A surgical airway was most often necessary when the injury involved Zone 1.
Facial wounds were most commonly located in Zone 2; 85 patients (12%) had wounds in more than one zone (Table 2). Open fractures or wounds that communicated with either the nasal sinuses or the oral cavity were present in 449 patients (62%) and were considered contaminated.
Urgent control of hemorrhage from the facial structures was necessary in 51 patients (7%). Control of bleeding was obtained by balloon tamponade in 3 (0.4%), angioembolization in 12 (2%), and anterior and posterior nasal packing in 25 (4%) and 11 (2%), respectively.
Following initial evaluation and resuscitation by the trauma service, definitive care was provided by multiple subspecialties in 271 patients (38%), solely trauma surgeons in 98 (14%), plastic surgeons in 79 (11%), neurosurgeons in 73 (10%), oral and maxillofacial surgeons in 92 (13%), ophthalmologists in 41 (6%), and otolaryngologists in 54 (8%).
Within 48 hours of admission, 239 patients (33%) achieved a disposition: death in 146 (20%) and either discharge or transfer (per trauma center–Health Maintenance Organization insurance contract provision) of stable patients in 105 (15%). Patients who died, who were discharged, or who were transferred within 48 hours were not included in the analysis of complications.
Of the 481 remaining patients, 74 received no operative procedures (15%); 16 of these patients were placed on comfort care because of associated severe brain injury. In the 407 patients requiring surgery, more than one operative procedure was performed on 355 patients (87%) and 182 (45%) had four or more procedures (Table 3). The most frequent type of operation involved the skin and soft tissue, consisting primarily of debridement and wound closure (e.g., skin grafting, flap coverage). The vast majority of patients requiring operative intervention received perioperative antibiotics (382, 94%).
Of the 481 patients remaining hospitalized for greater than 48 hours, a total of 207 complications developed in 184 patients (38%). Of these 207 complications, 96 (46%) were a complication of injury and 111 (54%) were a complication of care. Complications of injury included blindness, cranial nerve deficit, and cerebrovascular accident; complications of injury included infections, flap failure, and malunion or nonunion. The single most frequent complication was blindness, followed by cranial nerve deficit (Table 4). Infection or a complication of infection (e.g., pneumonia or malunion) occurred in a total of 67 patients (14%). Antibiotics (broad or narrow spectrum) were given to 447 patients (93%) within 24 hours of admission. There was no difference in the rate of infection between those who received antibiotics and those who did not (11% vs. 14%, p = 0.636). There was no difference in infection rate between those who received broad- versus narrow-spectrum antibiotics. However, only 21 patients did not receive antibiotics, and only 27 patients received what could be considered narrow-spectrum antibiotics.
Of the entire cohort of 720 patients, death occurred in 185 (26%). Brain injury was the major cause of death (180, 97%), two patients died of exsanguination, and one each of stroke (from an associated carotid artery injury), myocardial infarction, and multiple-organ failure.
We analyzed outcomes for isolated Zone I, II, and III injuries in the 481 patients hospitalized for greater than 48 hours. The mortality rates were significantly different by injuries to Zones I, II, and III, at 5.3%, 1.8%, and 26.2%, respectively (p < 0.001, data not shown). In-hospital complication rates also significantly differed by injury to Zones I, II, and III at 36%, 34%, and 14%, respectively (p < 0.01, data not shown). These relationships persisted in analyses of nonisolated zone injuries with regard to mortality (Zone I, 15.9%; Zone 2, 13.7%; and Zone 3, 59.3%; p < 0.001; data not shown) and complications (Zone I, 17.6%; Zone II, 6.6%; and Zone III, 10.1%; p < 0.05; data not shown).
Multivariable mixed model logistic regression was performed on the outcomes of in-hospital mortality, all combined complications, complications of injury, and complications of care (Table 5). All models were analyzed using the variables of patient age, sex, head Abbreviated Injury Scale (AIS) score, face AIS score, total operations performed, sources of bacterial contamination (nasal sinus or oral involvement), and Revised Trauma Score (RTS). Age, head AIS score, total operations, and RTS were significantly associated with mortality (all p < 0.05). Nasal and/or oral involvement and face AIS score were removed from this model because they were not related to mortality after adjustment. Increasing head AIS score (odds ratio [OR] 1.21; 95% confidence interval [CI], 1.05–1.39; p = 0.009), increasing number of total operations (OR, 1.14; 95% CI, 1.07–1.21; p < 0.01), and decreasing RTS (OR, 0.82; 95% CI, 0.73–0.93; p = 0.001) were significantly related to the development of complications. Face AIS score (OR, 2.16; 95% CI, 1.44–3.34; p < 0.001), oral involvement (OR, 0.27; 95% CI, 0.10–0.78; p = 0.016), and decreasing RTS (OR, 0.85; 95% CI, 0.10–0.78; p = 0.026) were significantly associated with complications of injury. Male sex (OR, 0.43; 95% CI, 0.21–0.88; p = 0.021), increasing head AIS score (OR, 1.19; 95% CI, 1.01–1.39; p = 0.034), increasing number of total operations (OR, 1.21; 95% CI, 1.12–1.29; p < 0.001), nasal sinus and oral involvement (OR, 2.93; 95% CI, 1.08–7.99; p = 0.035), and decreasing RTS (OR, 0.48; 95% CI, 0.39–0.60; p < 0.001) were significantly related to complications of care.
This report is the first to analyze the factors associated with adverse outcomes following gunshot wounds and blast injuries of the face. Mortality in our series is high at 26%, primarily owing to associated brain injury. Complications occurred in 38% of the patients, which seemed to be related to the admission physiology (low RTS) and the severity of injury. The association of patient complications with increasing operations is likely explained by the need for repeated operations in patients with complex anatomic injuries.
Our mortality rate of 26% is approximately twofold to fourfold that reported in the recent literature1,2,15 and is likely caused by our inclusion of the supraorbital area (frontal bone of the cranium) in our definition of the face. An injury to the supraorbital area was associated with a mortality of 59%, which is similar to the mortality rate of 51% for craniocerebral gunshot wounds reported by Murano et al.19 Our finding that brain injury was the cause of death in 97% of the patients would support this assumption. However, many Zone 1 and 2 injuries also resulted in a death caused by brain injury. Demetriades etal.1 and Dolin et al.,15 who did not include the supraorbital area, found significant brain injury in 49% and 17%, respectively. This supports their recommendation to use computed tomographic scanning of the head liberally in patients with gunshot wounds to the face. They also reported brain injury as a major cause of death.
Complications occurred in 38% of our patients, which is similar to the 25% to 85 % previously reported.4,8,15,20,21 The wide range of complication rates reported is a function of whether the author included only those caused by the injury itself (e.g., blindness from a direct injury to the globe) and excluded those occurring during the hospitalization (e.g., infection or venous thromboembolism). We included all complications. Similar to previous reports,2,8,21 the most common complications following this type of injury were ocular and neurologic—blindness, central nervous system or peripheral nerve deficit—and infection.
Management of these complex injuries requires multidisciplinary input and multiple operative procedures.1,8,15,21 Taher,8 in his detailed report of 1,135 patients with gunshot and blast injuries of the face, reported an average of 1.5 facial operations per patient. We report a much higher average of 5.4 operations per patient because we included all types of facial and extrafacial operations. If we included only those operations on the facial structures as reported by Taher,8 the operations per patient ranged between 1.3 and 2.0.
Early management of GSWBIF has been described by several authors who emphasize control of the airway and control of any obvious hemorrhage.1,4,15 Early airway control was necessary in 33%, which is similar to that reported by Dolin et al.15 However, in the majority of cases requiring emergent airway control, endotracheal intubation was successful. Surgical airways were necessary in only 5% of the patients requiring urgent airway control in this series. This parallels the experience of others who have analyzed the management of severe maxillofacial injuries. Early hemorrhage control was necessary in 7%, which is similar to that reported by Demetriades et al.1
The predominance of young male patients with this type of injury and the frequent use of the handgun are well known,1,4,8,15 but the frequency of comorbidities and positive toxicology findings have not been previously described. Medical comorbidities were present in 14%, and 10% had mental disorders documented in the medical record. A relatively high percentage of patients had positive toxicology findings, with illicit drugs present in 16% of the patients and alcohol present in 31%.
There are a number of limitations to this work. First, the investigation was retrospective and relied heavily on chart review and previous documentation. Second, it was a multicenter study, which introduced variation in the management and variation among chart abstractors. Third, because of the lack of follow-up information, we were unable to document the long-term morbidity. Fourth, measurement bias may have resulted because complications were evaluated only in survivors. Finally, the study covered an 11-year period during which time processes of care may have changed.
Despite these limitations, we conclude that GSWBIFs have high morbidity and mortality. The most common cause of death is brain injury. The need for an urgent surgical airway is infrequent. These injuries require multidisciplinary management and significant resource allocation with frequent returns to the operating room.
Further investigation is warranted. A standardized definition of the face is necessary to enhance comparative effectiveness research. The use of antibiotics must be clarified for these injuries because there is still a debate about the spectrum of coverage necessary and the duration of treatment.22,23 Postdischarge follow-up is necessary to determine both the long-term cosmetic outcome and the long-term morbidity of these devastating injuries. Future investigations should also note the presence of substance abuse and mental disorders when gathering data on this morbid and complex injury because these may be contributing factors to both the etiology and the outcome.
S.R.S. designed this study. S.R.S. and M.C.S. conducted the literature search. Individuals who collected, processed, and submitted center data included R.A.K., C.E.H., K.L.K., M.W., B.M.T., S.M.M., A.G.R., C.S.L., M.C.S., J.E.K., C.C.B., E.E.M., T.H.C., K.J.K., J.M.H., and J.W. J.E.K., M.C.S., and S.R.S. managed and cleaned the data. J.E.K. and R.Y.C. performed the statistical analyses. J.E.K., S.R.S., and R.Y.C. interpreted the data. S.R.S. and J.E.K. wrote the manuscript. J.E.K. prepared the tables and figure. S.R.S., J.E.K., C.C.B., E.E.M., R.A.K., T.H.C., K.L.K., C.S.L., B.M.T., A.G.R., and J.M.H. edited the final article.
We acknowledge the assistance of LCDR Daniel Witcher, DC, USN, in the design of the study and Beth Sise, MSN, in the editing and preparation of the manuscript.
The authors declare no conflicts of interest.
1. Demetriades D, Chahwan S, Gomez H, Falabella A, Velmahos G, Yamashita D . Initial evaluation and management of gunshot wounds to the face. J Trauma. 1998; 45: 39–41.
2. Hollier L, Grantcharova EP, Kattash M . Facial gunshot wounds: a 4-year experience. J Oral Maxillofac Surg. 2001; 59: 277–282.
3. Newlands SD, Samudrala S, Katzenmeyer WK . Surgical treatment of gunshot injuries to the mandible. Otolaryngol Head Neck Surg. 2003; 129: 239–244.
4. Neupert EA 3rd, Boyd SB . Retrospective analysis of low-velocity gunshot wounds to the mandible. Oral Surg Oral Med Oral Pathol. 1991; 72: 383–387.
5. Motamedi MH . Primary management of maxillofacial hard and soft tissue gunshot and shrapnel injuries. J Oral Maxillofac Surg. 2003; 61: 1390–1398.
6. Gant TD, Epstein LI . Low-velocity gunshot wounds to the maxillofacial complex. J Trauma. 1979; 19: 674–677.
7. Gussack GS, Jurkovich GJ . Penetrating facial trauma: a management plan. South Med J. 1988; 81: 297–302.
8. Taher AA . Management of weapon injuries to the craniofacial skeleton. J Craniofac Surg. 1998; 9: 371–382.
9. Prendergast PM . Anatomy of the Face and Neck. In: Shiffman MA, Di Giuseppe A. , eds. Cosmetic Surgery. Berlin, Germany: Springer Berlin; 2012; : 29–45.
10. Yao ST, Vanecko RM, Corley RD, Stuteville OH, Shoemaker WC . Gunshot wounds of the face. J Trauma. 1972; 12: 523–528.
11. Shuck LW, Orgel MG, Vogel AV . Self-inflicted gunshot wounds to the face: a review of 18 cases. J Trauma. 1980; 20: 370–377.
12. Kihtir T, Ivatury RR, Simon RJ, Nassoura Z, Leban S . Early management of civilian gunshot wounds to the face. J Trauma. 1993; 35: 569–577.
13. Glapa M, Kourie JF, Doll D, Degiannis E . Early management of gunshot injuries to the face in civilian practice. World J Surg. 2007; 31: 2104–2110.
14. Broadbent TR, Woolf RM . Gunshot wounds to the face: initial care. JTrauma. 1972; 12: 229–232.
15. Dolin J, Scalea T, Mannor L, Sclafani S, Trooskin S . The management of gunshot wounds to the face. J Trauma. 1992; 33: 508–515.
16. Owens BD, Kragh JF Jr, Wenke JC, Macaitis J, Wade CE, Holcomb JB . Combat wounds in Operation Iraqi Freedom and Operation Enduring Freedom. J Trauma. 2008; 64: 295–299.
17. Chan RK, Siller-Jackson A, Verrett AJ, Wu J, Hale RG . Ten years of war: a characterization of craniomaxillofacial injuries incurred during operations Enduring Freedom and Iraqi Freedom. J Trauma Acute Care Surg. 2012; 73:(6 Suppl 5): S453–S458.
18. Charlson ME, Pompei P, Ales KL, MacKenzie CR . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987; 40: 373–383.
19. Murano T, Mohr AM, Lavery RF, Lynch C, Homnick AT, Livingston DH . Civilian craniocerebral gunshot wounds: an update in predicting outcomes. Am Surg. 2005; 71: 1009–1014.
20. Zaytoun GM, Shikhani AH, Salman SD . Head and neck war injuries: 10-year experience at the American University of Beirut Medical Center. Laryngoscope. 1986; 96: 899–903.
21. Cole RD, Browne JD, Phipps CD . Gunshot wounds to the mandible and midface: evaluation, treatment, and avoidance of complications. Otolaryngol Head Neck Surg. 1994; 111: 739–745.
22. Lovato C, Wagner JD . Infection rates following perioperative prophylactic antibiotics versus postoperative extended regimen prophylactic antibiotics in surgical management of mandibular fractures. J Oral Maxillofac Surg. 2009; 67: 827–832.
23. Petersen K, Colyer MH, Hayes DK, Hale RG, Bell RB. Prevention of Combat-Related Infections Guidelines Panel. Prevention of infections associated with combat-related eye, maxillofacial, and neck injuries. JTrauma. 2011; 71:(2 Suppl 2): S264–S269.
Gunshot wounds to the face; outcomes; management
Copyright © 2014 by Lippincott Williams & Wilkins
Highlight selected keywords in the article text.