TRAUMA-ADULT RESPIRATORY DISTRESS SYNDROME Mark L. Walker, MD, FACS Atlanta, Georgia
Sixteen trauma victims with adult respiratory distress syndrome were retrospectively examined. High injury severity score (mean: 44), massive transfusion requirements, and prolonged ventilator days characterize this group. Persistent intra-abdominal infection accounted for two of three deaths in this series. Most of these patients were managed without paralysis using intermittent mandatory ventilation and positive-end expiratory pressure (PEEP). High frequency jet ventilation was necessary in one subject. Two patients exhibited early ARDS reversal, ie, clinical improvement, better chest x-ray and decreased shunt (within 72 hours) when an intra-abdominal septic focus was eradicated. Principles of critical care for these patients remain: an FiO2 < .40, limiting barotrauma, using PEEP for alveolar recruitment and close monitoring of 02 transport. (J Nati Med Assoc. 1991 ;83:501-504.) Key words * trauma * adult respiratory distress syndrome * intra-abdominal infection
From the Department of Surgery, Morehouse School of Medicine, Atlanta, Georgia. Presented at the 94th Annual Convention and Scientific Assembly of the National Medical Association, Orlando, Florida, July 15-20, 1989. Requests for reprints should be addressed to Dr Mark L. Walker, Department of Surgery, Morehouse School of Medicine, 35 Butler St, SE, Atlanta, GA 30335. JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
Adult respiratory distress syndrome (ARDS) remains a major cause of morbidity and mortality for the critically ill surgical patient. This is particularly the case in the setting of multi-organ failure. ' Although a substantial amount of data regarding ARDS in trauma victims exists, we have noted a few peculiar features of the syndrome in a small group of patients. This article shares this experience. The records of 16 trauma victims with ARDS managed by one surgeon over a 6-year period were reviewed. These cases were abstracted from a pool of 300 trauma victims. There were 11 males and five females. Adult respiratory distress syndrome was defined by: * PaO2/FiO2 S 250:1, * decreased pulmonary compliance < 40, * pulmonary capillary wedge pressure (PCWP) 18, and * diffuse chest x-ray (CXR) infiltrates. The mean injury severity score (ISS) for the group was 44. Almost three organ systems were injured on average. The blood transfusion requirements were massive (mean: 33.4 units/patient). However, four patients accounted for 60% of the transfusions. Excluding those four patients, 11.8 units of blood and blood products were transfused per patient. These patients spent on average 30 days on some type of ventilatory support, including continuous positive airway pressure (CPAP). Almost two thirds of their hospital days were spent on antibiotic medication, with an average of 7.6 antibiotics used per patient. A complete description of -
501
TRAUMA & ARDS
TABLE 1. TRAUMA-ADULT RESPIRATORY DISTRESS SYNDROME GROUP CHARACTERISTICS Mean Range Characteristic 17-73 42.3 Age 20-75 44.2 Injury severity score 11.4 4-15 Glasgow coma scale 2.76 1-5 Organ-systems injured 0-92 33.4 Blood transfusion 14-143 55.0 Hospital days 18-82 30.0 Ventilator days 14.5 10-18 Pulmonary capillary wedge pressure 5-30 15.0 CM Positive-end expiratory pressure
the group's characteristics is given in Table 1. Infection was common in this group; Table 2 lists the group's infections. One half of the group developed pneumonia. Seven of these patients had pulmonary contusion as one of their initial injuries. Three aspirated at some point after injury, one in the field and two during their intensive care unit stay; all survived. Nine were intubated in the emergency room, three in the field, and four in the operating room. Three were subjected to bronchoscopy because of mucous plugging or aspiration. In each instance, there was agreement between the trach aspirate and bronchial washing cultures. One postbronchial bacteremia was suspected but never confirmed. There was one death in the pneumonia group. This was an elderly (72 years old) female with coronary artery disease, cerebral and pulmonary contusions, a left supracondylar femur fracture, an open right tibia-fibula fracture, and a left proximal radius fracture. The patient developed myocardial infarction shock during weaning while on CPAP. She had autopsy-proven ARDS and broncho-
pneumonia. Intra-abdominal infection occurred in five patients; three had enterococcal bacteremia. Group D enterococcus was the most common organism cultured from peritoneal fluid in this group. Three patients were packed during the initial exploratory laparotomy because of coagulopathy, serious liver injuries, and hypothermia; two of the three died. Both had hepatic artery ligation as part of their operative treatment, and both developed patchy necrosis of the right lobe necessitating right hepatic lobectomy. One died with Pseudomonas cepacia bacteremia and septic shock, and the other died from disseminated intravascular coagulation and overwhelming postsplenectomy sepsis. If right hepatic artery ligation is required initially for control of exsanguinating hemorrhage along with liver packing, 502
TABLE 2. INFECTIONS IN TRAUMA-ADULT RESPIRATORY DISTRESS SYNDROME No. of Patients Mortality % Infection 12.5 1 8 Pneumonia (S aureus) 40 2 5 Intra-abdominal (Group D enterococcus) 0 2 0 Line-related sepsis 0 0 5 Urinary tract infection
formal right hepatic lobectomy is recommended at the time the packs are removed. Catheter sepsis was documented in two patients. One elderly farmer with flail chest and pulmonary contusion developed Staphylococcus epidermidis bacteremia after a central line (previously changed over a guide wire) stayed in for 11 days. The second patient sustained massive hemorrhage from 15 stab wounds to the trunk, flank, and neck. A central line placed in the operating room and left in for 15 days was the culprit. The message here is not new. Lines inserted in the emergency room or operating room must be considered contaminated and removed within the first 24 hours. Twelve of 16 patients (75% of this trauma ARDS group) were bacteremic during their hospital course. There were 3.5 bacteremia days on average for these 12 patients. Staphylococcus species accounted for most of these episodes (13 episodes in nine patients). Enterobacter species, Candida species, and group D enterococcus comprised the top four organisms causing bacteremia. Polymicrobial bacteremia occurred once and was indicative of a persistent intra-abdominal infection requiring operation. Ventilatory management was straightforward in this group. Most patients received mechanical ventilatory support using intermittent mandatory ventilation (IMV) and positive-end expiratory pressure (PEEP). Two received continuous PEEP, two received CPAP only, and one required high frequency jet ventilation. Peak inspiratory pressure of the latter reached 100 cm H20, and significant air leaks from the chest tubes were present. High frequency jet ventilation was effective in lowering mean airway pressure, and the patient survived. Vasodilators, inotropes, and diuretics were used often. Eleven of 16 patients received diuretics during their treatment course. Ten were maintained on dopamJOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
TRAUMA & ARDS
ine, and two received dobutamine. Four patients required nitroprusside. Although pulmonary artery pressures were controlled in each instance, nitroprusside contributed to hypoxemia in two cases and had to be discontinued. Presumably, this occurred because it obliterated protective pulmonary hypoxic vasoconstriction, thus increasing shunt flow. In two patients, dramatic improvement in ARDS occurred with a decrease in shunt, an increase in compliance, and a decrease in ventilatory support after definitive surgical treatment. One patient with an intra-abdominal abscess status postdistal pancreatectomy and splenectomy was on 30 cm PEEP preoperatively. Seventy-two hours after the abscess was drained, the patient was on 10 cm CPAP. Another had dramatic improvement in his chest x-ray and clinical picture after debridement of residual necrotic flank musculature from a self-inflicted shotgun wound. This type of reversal is seen early in the course of ARDS (usually during the first week, provided the septic focus is properly eradicated). Survivors were compared with nonsurvivors (Table 3). No firm statistical analyses are possible because of the small numbers. However, the nonsurvivors received more blood, had more bacteremias, and had more operations than the survivors. Oxygen delivery was optimized using Shoemaker's criteria 800 mL/minute m2 in four patients. Three of this group survived.
DISCUSSION In the trauma setting, ARDS is often related to severe multisystem injury, the need for massive transfusion, the presence of pulmonary contusion and intraabdominal infection. It is an accepted axiom that those patients with established ARDS who become septic have pneumonia. Those who are septic and develop ARDS usually have an intra-abdominal infection that needs operative treatment. Most of the patients in this study were managed with routine ventilatory support techniques (eg, IMV and PEEP). Intermittent mandatory ventilation provides a set number of machine breaths per minute but allows spontaneous breathing to take place. Mean airway pressure is reduced, and barotrauma is limited. Increased work of breathing may be seen during weaning because of the demand valve setup in most systems. Pressure support can minimize this problem. We recommend an FiO2 .40 RR 10 TV = 10 mL/kg and PEEP > 5 for these patients initially. High frequency jet ventilation is a new technique that achieves adequate CO2 removal at rapid rates (80 to -
,
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
TABLE 3. SURVIVORS VERSUS NONSURVIVORS ADULT RESPIRATORY DISTRESS SYNDROME Survivors Nonsurvivors 38.0 39.0 Age 40.0 44.0 ISS 2.66 2.8 Average no. of organ systems injured 4.6 2.6 No. of operations 60.0 26.0 No. of transfusions 36.0 28.4 Ventilator days 5.06 2.08 No. of bacteremias 1.0 3.0 02 delivery optimized 2.0 10.0 02 delivery not
optimized
400) with small tidal volumes, 1 mL/kg to 5 mL/kg. It is especially useful when peak airway pressures are enormous, or in the setting of a major bronchopleural fistula.2 A special endotracheal tube is required, and careful attention must be paid to humidification as the tracheobronchial tree can dry out and erosions from the pulsatile flow can occur. No modality of ventilatory support changes the prognosis in ARDS. Positive-end expiratory pressure does not influence the course of the syndrome nor does it prevent ARDS.3 Positive-end expiratory pressure must not be withdrawn too rapidly as alveolar recruitment may be lost and atelectasis with hypoxemia may result. We recommend 2.5-cm drops over 2-hour periods or longer as improvement occurs. As PEEP is increased, adequate volume must be maintained or cardiac output will fall. If filling pressures are optimal, inotropes may be added to maintain flow. Oxygen transport assessment (including 02 extraction ratios and mixed venous 02 saturation) is essential. We note that acute lung injury (early ARDS) can be rapidly reversed daring the first week if a septic intra-abdominal focus is the cause and handled appropriately. Two patients exhibited dramatic improvement in the first week of their course. Recent work detailing elevated serum levels of tumor necrosis factor in a subset of septic patients may have some bearing on this.4 We hypothesize that as inflammatory mediator levels fall rapidly with obliteration of the septic focus, the capillary leak stops. If the pulmonary lymphatics and interstitium are for the most part intact, quick resolution of the pulmonary edema should occur. Later on in the course (second through third week), more gradual improvement is the rule. Infections remain the major problem in trauma-ARDS patients. Staphylococcus species remain the major cause of pneumonia, bacteremia, and line sepsis in these 503
TRAUMA & ARDS
seriously ill patients. Others have noted this pattern in multi-organ failure patients.4 Obliteration of normal flora by antibiotics, bacterial adherence to epithelial cells, and slime production are all factors that play some role.5 We use broad spectrum coverage for the ARDS patient with sepsis, emphasizing adequate staphylococcal coverage (usually vancomycin). Surveillance sputum monitoring is helpful in selecting antibiotics. With aggressive treatment of infection and careful ventilatory management with 02 transport assessment, the outlook for these patients should be good. The use of thromboxane synthetase inhibitors with blood transfusions should be investigated. Free oxygen radical scavengers and tumor necrosis factor inhibitors represent a current wave in research and prophylaxis for ARDS.
504
Literature Cited 1. Pepe PE. The clinical entity of adult respiratory distress syndrome: definition prediction and prognosis. Crit Care Clin. 1 986;2:398-399. 2. Slader A. High Frequency Jet Ventilation in Trauma. Advances in Trauma. Chicago, III: Year Book Medical Publishers, 1986:168-191. 3. Ingbar DH, Matthay, RA. Pulmonary sequelae and lung repair in survivors of ARDS. Acute lung injury. Crit Care Clin. 1986;2:651. 4. Debets JMH, Kampmeijer R, van der Linden, et al. Plasma tumor necrosis factor and mortality in critically ill septic patients. Crit Care Med. 1 989;1 7:489-494. 5. Marshall JC, Christou NV, Horn R, et al. The microbiology of multiple organ failure: the proximal gastrointestinal tract as an occult reservoir of pathogens. Arch Surg. 1988;123:309315.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
Sixteen trauma victims with adult respiratory distress syndrome were retrospectively examined. High injury severity score (mean: 44), massive transfusion requirements, and prolonged ventilator days characterize this group. Persistent intra-abdominal infection accounted for two of three deaths in this series. Most of these patients were managed without paralysis using intermittent mandatory ventilation and positive-end expiratory pressure (PEEP). High frequency jet ventilation was necessary in one subject. Two patients exhibited early ARDS reversal, ie, clinical improvement, better chest x-ray and decreased shunt (within 72 hours) when an intra-abdominal septic focus was eradicated. Principles of critical care for these patients remain: an FiO2 < .40, limiting barotrauma, using PEEP for alveolar recruitment and close monitoring of 02 transport. (J Nati Med Assoc. 1991 ;83:501-504.) Key words * trauma * adult respiratory distress syndrome * intra-abdominal infection
From the Department of Surgery, Morehouse School of Medicine, Atlanta, Georgia. Presented at the 94th Annual Convention and Scientific Assembly of the National Medical Association, Orlando, Florida, July 15-20, 1989. Requests for reprints should be addressed to Dr Mark L. Walker, Department of Surgery, Morehouse School of Medicine, 35 Butler St, SE, Atlanta, GA 30335. JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
Adult respiratory distress syndrome (ARDS) remains a major cause of morbidity and mortality for the critically ill surgical patient. This is particularly the case in the setting of multi-organ failure. ' Although a substantial amount of data regarding ARDS in trauma victims exists, we have noted a few peculiar features of the syndrome in a small group of patients. This article shares this experience. The records of 16 trauma victims with ARDS managed by one surgeon over a 6-year period were reviewed. These cases were abstracted from a pool of 300 trauma victims. There were 11 males and five females. Adult respiratory distress syndrome was defined by: * PaO2/FiO2 S 250:1, * decreased pulmonary compliance < 40, * pulmonary capillary wedge pressure (PCWP) 18, and * diffuse chest x-ray (CXR) infiltrates. The mean injury severity score (ISS) for the group was 44. Almost three organ systems were injured on average. The blood transfusion requirements were massive (mean: 33.4 units/patient). However, four patients accounted for 60% of the transfusions. Excluding those four patients, 11.8 units of blood and blood products were transfused per patient. These patients spent on average 30 days on some type of ventilatory support, including continuous positive airway pressure (CPAP). Almost two thirds of their hospital days were spent on antibiotic medication, with an average of 7.6 antibiotics used per patient. A complete description of -
501
TRAUMA & ARDS
TABLE 1. TRAUMA-ADULT RESPIRATORY DISTRESS SYNDROME GROUP CHARACTERISTICS Mean Range Characteristic 17-73 42.3 Age 20-75 44.2 Injury severity score 11.4 4-15 Glasgow coma scale 2.76 1-5 Organ-systems injured 0-92 33.4 Blood transfusion 14-143 55.0 Hospital days 18-82 30.0 Ventilator days 14.5 10-18 Pulmonary capillary wedge pressure 5-30 15.0 CM Positive-end expiratory pressure
the group's characteristics is given in Table 1. Infection was common in this group; Table 2 lists the group's infections. One half of the group developed pneumonia. Seven of these patients had pulmonary contusion as one of their initial injuries. Three aspirated at some point after injury, one in the field and two during their intensive care unit stay; all survived. Nine were intubated in the emergency room, three in the field, and four in the operating room. Three were subjected to bronchoscopy because of mucous plugging or aspiration. In each instance, there was agreement between the trach aspirate and bronchial washing cultures. One postbronchial bacteremia was suspected but never confirmed. There was one death in the pneumonia group. This was an elderly (72 years old) female with coronary artery disease, cerebral and pulmonary contusions, a left supracondylar femur fracture, an open right tibia-fibula fracture, and a left proximal radius fracture. The patient developed myocardial infarction shock during weaning while on CPAP. She had autopsy-proven ARDS and broncho-
pneumonia. Intra-abdominal infection occurred in five patients; three had enterococcal bacteremia. Group D enterococcus was the most common organism cultured from peritoneal fluid in this group. Three patients were packed during the initial exploratory laparotomy because of coagulopathy, serious liver injuries, and hypothermia; two of the three died. Both had hepatic artery ligation as part of their operative treatment, and both developed patchy necrosis of the right lobe necessitating right hepatic lobectomy. One died with Pseudomonas cepacia bacteremia and septic shock, and the other died from disseminated intravascular coagulation and overwhelming postsplenectomy sepsis. If right hepatic artery ligation is required initially for control of exsanguinating hemorrhage along with liver packing, 502
TABLE 2. INFECTIONS IN TRAUMA-ADULT RESPIRATORY DISTRESS SYNDROME No. of Patients Mortality % Infection 12.5 1 8 Pneumonia (S aureus) 40 2 5 Intra-abdominal (Group D enterococcus) 0 2 0 Line-related sepsis 0 0 5 Urinary tract infection
formal right hepatic lobectomy is recommended at the time the packs are removed. Catheter sepsis was documented in two patients. One elderly farmer with flail chest and pulmonary contusion developed Staphylococcus epidermidis bacteremia after a central line (previously changed over a guide wire) stayed in for 11 days. The second patient sustained massive hemorrhage from 15 stab wounds to the trunk, flank, and neck. A central line placed in the operating room and left in for 15 days was the culprit. The message here is not new. Lines inserted in the emergency room or operating room must be considered contaminated and removed within the first 24 hours. Twelve of 16 patients (75% of this trauma ARDS group) were bacteremic during their hospital course. There were 3.5 bacteremia days on average for these 12 patients. Staphylococcus species accounted for most of these episodes (13 episodes in nine patients). Enterobacter species, Candida species, and group D enterococcus comprised the top four organisms causing bacteremia. Polymicrobial bacteremia occurred once and was indicative of a persistent intra-abdominal infection requiring operation. Ventilatory management was straightforward in this group. Most patients received mechanical ventilatory support using intermittent mandatory ventilation (IMV) and positive-end expiratory pressure (PEEP). Two received continuous PEEP, two received CPAP only, and one required high frequency jet ventilation. Peak inspiratory pressure of the latter reached 100 cm H20, and significant air leaks from the chest tubes were present. High frequency jet ventilation was effective in lowering mean airway pressure, and the patient survived. Vasodilators, inotropes, and diuretics were used often. Eleven of 16 patients received diuretics during their treatment course. Ten were maintained on dopamJOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
TRAUMA & ARDS
ine, and two received dobutamine. Four patients required nitroprusside. Although pulmonary artery pressures were controlled in each instance, nitroprusside contributed to hypoxemia in two cases and had to be discontinued. Presumably, this occurred because it obliterated protective pulmonary hypoxic vasoconstriction, thus increasing shunt flow. In two patients, dramatic improvement in ARDS occurred with a decrease in shunt, an increase in compliance, and a decrease in ventilatory support after definitive surgical treatment. One patient with an intra-abdominal abscess status postdistal pancreatectomy and splenectomy was on 30 cm PEEP preoperatively. Seventy-two hours after the abscess was drained, the patient was on 10 cm CPAP. Another had dramatic improvement in his chest x-ray and clinical picture after debridement of residual necrotic flank musculature from a self-inflicted shotgun wound. This type of reversal is seen early in the course of ARDS (usually during the first week, provided the septic focus is properly eradicated). Survivors were compared with nonsurvivors (Table 3). No firm statistical analyses are possible because of the small numbers. However, the nonsurvivors received more blood, had more bacteremias, and had more operations than the survivors. Oxygen delivery was optimized using Shoemaker's criteria 800 mL/minute m2 in four patients. Three of this group survived.
DISCUSSION In the trauma setting, ARDS is often related to severe multisystem injury, the need for massive transfusion, the presence of pulmonary contusion and intraabdominal infection. It is an accepted axiom that those patients with established ARDS who become septic have pneumonia. Those who are septic and develop ARDS usually have an intra-abdominal infection that needs operative treatment. Most of the patients in this study were managed with routine ventilatory support techniques (eg, IMV and PEEP). Intermittent mandatory ventilation provides a set number of machine breaths per minute but allows spontaneous breathing to take place. Mean airway pressure is reduced, and barotrauma is limited. Increased work of breathing may be seen during weaning because of the demand valve setup in most systems. Pressure support can minimize this problem. We recommend an FiO2 .40 RR 10 TV = 10 mL/kg and PEEP > 5 for these patients initially. High frequency jet ventilation is a new technique that achieves adequate CO2 removal at rapid rates (80 to -
,
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
TABLE 3. SURVIVORS VERSUS NONSURVIVORS ADULT RESPIRATORY DISTRESS SYNDROME Survivors Nonsurvivors 38.0 39.0 Age 40.0 44.0 ISS 2.66 2.8 Average no. of organ systems injured 4.6 2.6 No. of operations 60.0 26.0 No. of transfusions 36.0 28.4 Ventilator days 5.06 2.08 No. of bacteremias 1.0 3.0 02 delivery optimized 2.0 10.0 02 delivery not
optimized
400) with small tidal volumes, 1 mL/kg to 5 mL/kg. It is especially useful when peak airway pressures are enormous, or in the setting of a major bronchopleural fistula.2 A special endotracheal tube is required, and careful attention must be paid to humidification as the tracheobronchial tree can dry out and erosions from the pulsatile flow can occur. No modality of ventilatory support changes the prognosis in ARDS. Positive-end expiratory pressure does not influence the course of the syndrome nor does it prevent ARDS.3 Positive-end expiratory pressure must not be withdrawn too rapidly as alveolar recruitment may be lost and atelectasis with hypoxemia may result. We recommend 2.5-cm drops over 2-hour periods or longer as improvement occurs. As PEEP is increased, adequate volume must be maintained or cardiac output will fall. If filling pressures are optimal, inotropes may be added to maintain flow. Oxygen transport assessment (including 02 extraction ratios and mixed venous 02 saturation) is essential. We note that acute lung injury (early ARDS) can be rapidly reversed daring the first week if a septic intra-abdominal focus is the cause and handled appropriately. Two patients exhibited dramatic improvement in the first week of their course. Recent work detailing elevated serum levels of tumor necrosis factor in a subset of septic patients may have some bearing on this.4 We hypothesize that as inflammatory mediator levels fall rapidly with obliteration of the septic focus, the capillary leak stops. If the pulmonary lymphatics and interstitium are for the most part intact, quick resolution of the pulmonary edema should occur. Later on in the course (second through third week), more gradual improvement is the rule. Infections remain the major problem in trauma-ARDS patients. Staphylococcus species remain the major cause of pneumonia, bacteremia, and line sepsis in these 503
TRAUMA & ARDS
seriously ill patients. Others have noted this pattern in multi-organ failure patients.4 Obliteration of normal flora by antibiotics, bacterial adherence to epithelial cells, and slime production are all factors that play some role.5 We use broad spectrum coverage for the ARDS patient with sepsis, emphasizing adequate staphylococcal coverage (usually vancomycin). Surveillance sputum monitoring is helpful in selecting antibiotics. With aggressive treatment of infection and careful ventilatory management with 02 transport assessment, the outlook for these patients should be good. The use of thromboxane synthetase inhibitors with blood transfusions should be investigated. Free oxygen radical scavengers and tumor necrosis factor inhibitors represent a current wave in research and prophylaxis for ARDS.
504
Literature Cited 1. Pepe PE. The clinical entity of adult respiratory distress syndrome: definition prediction and prognosis. Crit Care Clin. 1 986;2:398-399. 2. Slader A. High Frequency Jet Ventilation in Trauma. Advances in Trauma. Chicago, III: Year Book Medical Publishers, 1986:168-191. 3. Ingbar DH, Matthay, RA. Pulmonary sequelae and lung repair in survivors of ARDS. Acute lung injury. Crit Care Clin. 1986;2:651. 4. Debets JMH, Kampmeijer R, van der Linden, et al. Plasma tumor necrosis factor and mortality in critically ill septic patients. Crit Care Med. 1 989;1 7:489-494. 5. Marshall JC, Christou NV, Horn R, et al. The microbiology of multiple organ failure: the proximal gastrointestinal tract as an occult reservoir of pathogens. Arch Surg. 1988;123:309315.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 83, NO. 6
