a b s t r a c t

Study objectives: The rate of iatrogenic pneumothorax associated with thoracic central venous catheterization in community emergency departments (EDs) is poorly described, although such information is vital to inform the procedure’s risk/benefit analysis. We undertook this multicenter study to estimate the incidence of immediate catheter-related pneumothorax in community EDs and to determine associations with site of access, failed ac- cess, and positive pressure ventilation.

Methods: This was a secondary analysis of 2 retrospective cohort studies of adults who underwent attempted tho- racic central venous catheterization in 1 of 21 EDs. Pneumothorax was identified by postprocedural anteroposterior chest radiograph or emergent evacuation for presumed Tension pneumothorax. Frequencies were compared using Fisher’s exact test.

Results: Among 1249 patient encounters, the initial vein of catheterization was internal jugular in 1054 cases (84.4%) and subclavian in 195 cases (15.6%). Success at the initial internal jugular vein was more common than at the initial subclavian vein (95.4% vs 83.6%, P b .001). Periprocedural positive pressure ventilation was ad- ministered in 316 patients (25.3%). We identified 6 pneumothoraces (0.5%; 95% confidence interval, 0.2%-1.1%). The Incidence of pneumothorax was higher with the subclavian vein than the internal jugular vein (2.3% vs 0.1%, P b .001), with failed access at the initial vein (2.5% vs 0.3%, P = .05), and among patients receiving positive pres- sure ventilation (1.6% vs 0.1%, P b .01).

Conclusion: The incidence of pneumothorax from Thoracic central venous catheterization in community EDs is low. The risk of pneumothorax is higher with a subclavian vein approach, failed access at the initial vein, and positive pressure ventilation.

(C) 2014

Introduction

Central venous catheterization can be a mainstay for the delivery of fluids and medications to patients without peripheral access and those who are critically ill or injured. The success and safety of the procedure have improved significantly in this era of ultrasound use [1,2]. Yet there remain valid concerns about mechanical, infectious, and thromboembolic complications. Among the most threatening mechanical complications

? Sources of support: The study was supported by the Kaiser Permanente Northern California Community Benefit Program, Oakland, CA, and the Gordon and Betty Moore Foundation, Palo Alto, CA.

* Corresponding author at: Department of Emergency Medicine, Kaiser Permanente Roseville Medical Center, 1600 Eureka Rd, Roseville, CA 95661. Tel.: +1 916 973 6901×1613.

E-mail address: [email protected] (D.R. Vinson).

that immediately attend thoracic central line placement is pneumotho- rax. Iatrogenic pneumothorax directly effects patient morbidity, as it often requires an evacuation procedure. Moreover, this complication is known to increase health care resource use and mortality. The US Agency for Healthcare Research and Quality found in 2012 that iatrogenic pneu- mothorax was associated with 4.4 days of extra hospitalization, more than $17,000 in additional hospital charges, and 7.0% excess mortality [3]. The rates of central venous catheterization in emergency medicine have been on the rise [4,5]. Nearly all emergency medicine research on central venous catheterization, however, has been performed in an academic setting [6-12]. Moreover, the bulk of the literature on the in- cidence of iatrogenic pneumothorax has examined the procedure in the hands of intensivists, anesthesiologists, and surgeons. Less is known about the safety of the procedure in the community emergency department (ED) setting, where the majority of emergency care in the

http://dx.doi.org/10.1016/j.ajem.2014.10.020

0735-6757/(C) 2014

United States is provided [13]. A more accurate estimation of these con- temporary complication rates is needed to help inform the risk/benefit calculus of this common procedure. This is particularly important be- cause emergency physicians report that the risk of iatrogenic complica- tions is an impediment to central venous catheterization [14]. In addition, informed consent undertaken by emergency physicians has resulted in as many as one-fourth of patients refusing central venous catheterization after hearing of the perceived procedural risks [15].

We undertook this secondary analysis of 2 retrospective cohort studies with 2 aims in mind: (1) to estimate the incidence of immediate iatrogenic pneumothorax following thoracic central venous catheteriza- tion in community EDs and (2) to determine the association between vein site (internal jugular vs subclavian), initial catheterization site fail- ure, and positive pressure ventilation on pneumothorax rates.

Methods

Study design and setting

This analysis includes adult (>=18 years) patients with sepsis under- going attempted thoracic central venous catheterization identified from 2 retrospective cohort studies [15,16]. These prior studies were under- taken in 21 EDs within Kaiser Permanente (KP) Northern California, a large integrated Health care delivery system that provides comprehen- sive care for more than 3.4 million members. The study facilities are nonrural community hospitals with inpatient bed capacities ranging from 50 to 325. The KP health system is supported by an Epic-based (Verona, WI) electronic health record implemented in 2005 [17].

KP Northern California EDs are staffed by approximately 500 residency-trained, board-certified (or board-eligible) emergency physi- cians. The mean annual census per department during 2011 was 43500 visits (range, 22000-78 000). Seven of the 21 EDs serve as satellite clinical rotation sites for university Emergency Medicine residency training pro- grams. One ED cohosts a residency training program with a university hospital. Thirteen EDs have no academic affiliation. Four medical centers had trauma center designation (level II or lower) at the time of the study. The study period followed the implementation of a protocol-driven approach to sepsis management that was part of a system-wide quality improvement initiative that included a training program at each facility on sepsis diagnoses, management, and ultrasound-guided thoracic cen- tral venous catheterization [15,16,18]. At the time of the study, all EDs had at least 1 ultrasound machine within the department for emergen- cy physician use. The predominant catheterization device in use during the study period across all EDs was a triple-lumen central venous cath- eter with capability for continuously monitoring Central venous oxygen saturation (PreSep Oximetry catheters; Edwards Lifesciences Corpora- tion, Irvine, CA). The other components of our medical group‘s perfor- mance improvement program have been previously reported [18]. Proceduralists used a standard modified Seldinger technique [19]. All patients undergoing attempted thoracic central venous catheterization in our EDs received postprocedural chest radiography to assess for line placement and pneumothorax. The KP Northern California Health Ser-

vices Institutional Review Board approved this study.

Selection of participants

Eligible cases were identified from 2 prior retrospective cohort stud- ies. The first of these studies identified variables predictive of unattempted central venous catheterization in ED patients with severe sepsis or septic shock who met criteria for protocolized care between August 1, 2009, and August 31, 2010, in 5 community EDs in KP Northern California [15]. The second study quantified the rates of hemor- rhagic complications of central line placement in septic patients with at least one hemostatic laboratory abnormality between March 1, 2010, and June 30, 2012, in 21 community EDs [16]. The primary outcome of in- terest for this present study, immediate pneumothorax after attempted

thoracic central venous catheterization, was included in the initial data abstraction objectives of the 2 initial studies.

Our present study cohort is the population of ED patients enrolled in these 2 study cohorts who underwent attempted thoracic central ve- nous catheterization. Patients that were eligible for study inclusion on different occasions (that is, during different patient encounters on dif- ferent dates) were enrolled more than once.

Methods and measurements

Initial studies

The general data collection methods and the variables identified for the 2 initial studies have been previously described [15,16]. Abstractors for these initial studies were blind to this study’s secondary aim of de- termining the associations of vein approach, failed access, and positive pressure ventilation with the incidence of pneumothorax. Abstractors were informed only of the aims of the initial 2 studies, which addressed rates and predictors of unattempted Central lines and line-related hem- orrhagic complications, respectively [15,16].

The current study

For this present study, we undertook additional data collection pertaining to periprocedural positive pressure ventilation. We electron- ically identified from the health care system’s comprehensive adminis- trative databases which patients in our cohort had received positive pressure ventilation during their ED stay, including noninvasive positive pressure ventilation (continuous positive airway pressure ventilation and bilevel positive airway pressure ventilation) and invasive positive pressure ventilation (ie, mechanical ventilation). We then manually reviewed the electronic health records of each of these cases to ascertain the temporal relationship of the positive pressure ventilation with the attempted thoracic central venous catheterization(s). Noninvasive pos- itive pressure ventilation was categorized as occurring only before (but not during or after), before and after (and maybe during), and only after the central venous catheterization. Noninvasive positive pressure ventila- tion that occurred before (but not during or after) the vein catheterization was not considered periprocedural positive pressure ventilation for the purposes of this study and is not reported. For patients receiving noninva- sive positive pressure ventilation both before and after attempted central venous catheterization, we sought to determine if the positive pressure ventilation was continued or interrupted during the catheterization at- tempt. Invasive positive pressure ventilation was categorized as having been initiated before or after the vein catheterization.

Our primary outcome was a postprocedural pneumothorax. An im- mediate pneumothorax was defined as an appropriately sided collec- tion of air in the pleural space identified on the initial postprocedural anteroposterior chest radiograph (supine or upright) in a board- certified radiologist’s official interpretation or by emergent evacuation of a presumed tension pneumothorax. Subsequent chest films were also reviewed for patients who underwent further ED or postintubation imaging. It is known that anteroposterior chest radiography is not the most sensitive test for pneumothorax [20]. However, because it is the standard Diagnostic modality widely used for this purpose both in emergency clinical practice and in clinical research, it is the diagnostic method we used.

Additional data collection was undertaken for this study with regard to the cases of pneumothorax. To ensure that we identified all cases of immediate postcatheterization pneumothorax in our cohort, we searched the health care system’s administrative databases for a dis- charge diagnosis of iatrogenic pneumothorax from the index hospitali- zation using International Classification of Diseases, Ninth Revision (ICD- 9) code 512.1. An ICD-9 code in isolation has been shown to be specific but insensitive for the detection of complications of thoracic central venous catheterization, which is why we used it as an adjunct to our comprehensive manual chart review [21]. Cases identified with a pneu- mothorax diagnosis on hospital discharge prompted a manual review of

their electronic health records, including physician notes and Radiology reports, to determine if the patients had an iatrogenic pneumothorax from an attempted thoracic central line during their ED stay that had

Table 1

Initial and second central vein catheterization attempts

N = 1249 patient encounters

been missed on initial data abstraction.

All cases of postprocedural pneumothorax underwent additional manual chart review by a co-investigator (VAC, a board-certified radiol- ogist) to report the nature of the diagnostic chest radiograph (supine or upright) and to measure the size of the pneumothorax, reported as a percentage. This was calculated using a formula that accounts for whether the examination was performed supine or upright and assesses the average Interpleural distance based on 3 measurements from the chest wall to the pleural margin [22,23]. We also report the presence and degree of mediastinal shift (absent, small, moderate, or large). In addition, we determined the nature of the therapeutic intervention (Pigtail catheter with Heimlich valve or tube thoracostomy) and de- scribe the duration and outcome of the drainage treatment.

Five percent of all cases (n = 65) were randomly selected for indepen- dent review by a second investigator to assess for interrater reliability on site of catheterization, success of catheterization, and pneumothorax out- come, reported as percentage of agreement.

Statistical analysis

Continuous variables are presented as means with standard devia- tion, and categorical data are presented as the percentage of frequency of occurrence. Frequencies observed between groups were compared using Fisher’s exact test. Confidence intervals (CIs) were calculated using the modified Wald method. A 2-tailed P value of less than .05 was considered to indicate statistical significance.

Results

Our cohort was comprised of 1229 patients, 20 of whom were eligi- ble for the study on 2 separate occasions, bringing the total number of eligible patient encounters (or cases) to 1249. The mean age of the co- hort was 67.5 years (+- 15.2); 543 patients (44.2%) were female. The proceduralists were predominantly attendings (n = 1083, 86.7%).

The first site of attempted thoracic central venous access was inter- nal jugular in 1054 cases (84.4%) and subclavian in 195 cases (15.6%). The approach to the latter (infraclavicular vs supraclavicular) was rarely documented, but nearly all of our physicians report undertaking the more common infraclavicular approach: a large survey found that only 2% of emergency physicians in these 21 EDs performed more than 2 supraclavicular subclavian catheterizations per year [14]. The right side was strongly favored over the left for both the initial internal jugular (n = 965, 91.6%) and the initial subclavian (n = 150, 76.9%) vein catheterization attempts.

The success rate at the first thoracic vein site was 93.5% (n = 1168). Success at the initial internal jugular vein was more common than at the initial subclavian vein (95.4% vs 83.6%, P b .001) (Table 1). Eighty-one cases (6.5%) experienced a failed catheterization at one or more central vein sites. Of these 81 cases, 77 underwent attempted catheterization at a second site, 13 of which failed (13/77, 16.9%). These 13 cases all underwent attempted catheterization at a third site, only one of which failed. Overall, 44 cases (3.5%) underwent attempts at both internal jug- ular and subclavian veins (41 switched veins for the second attempt and 3 for the third attempt). Table 1 describes the location of the thoracic central vein catheterization attempts.

Counting each central vein catheterization site as a separate attempt, there were 1347 total catheterization attempts, 1327 of which were thoracic in location. We report the incidence per patient encounter (or case, n = 1249) and not per thoracic central venous catheterization attempt (n = 1327).

We identified 6 pneumothoraces (0.5%; 95% CI, 0.2%-1.1%). All 6 were detected by both manual chart review and electronic ICD-9 code search, which was 100% sensitive. No additional pneumothoraces

Thoracic central vein site n (%)

Initial internal jugular 1054 (84.4)

Success 1005 (95.4)

Failure 49 (4.6)

Subsequent postjugular site

Contralateral internal jugular	16
Subclavian	22
Femoral	8
None	3
Initial subclavian	195 (15.6)
Success	163 (83.6)
Failure	32 (16.4)
Subsequent postsubclavian site Contralateral subclavian Internal jugular	6 19
Femoral	6
None	1

were identified from the ICD-9 code search that had not already been identified by manual chart review. All cases of pneumothorax were de- tected on postcatheterization chest radiograph. None had received emer- gency preradiographic drainage for presumed tension pneumothorax. The proceduralists were ED attendings in all 6 cases. The size of the pneumothoraces and degree of mediastinal shift are reported in Table 2. Five of the 6 pneumothoraces occurred following attempted subcla- vian vein catheterization, 4 successful and 1 unsuccessful (Table 2). The unsuccessful subclavian vein catheterization was followed by a success- ful ultrasound-guided internal jugular catheterization on the same side prior to the radiographic diagnosis of pneumothorax. One patient devel- oped a pneumothorax following an isolated failed internal jugular vein catheterization. When stratified by vein, the incidence of pneumothorax was significantly higher in cases who had received an attempted subcla- vian vein catheterization than in those who received only an attempted

internal jugular vein catheterization (2.3% vs 0.1%, P b .001, Table 3).

The incidence of immediate pneumothorax was higher in cases in which the initial catheterization was unsuccessful compared with their successful counterparts (2.5% vs 0.3%), but the difference did not quite reach statistical significance (P = .05) (Table 3).

Of the 1249 cases, 316 (25.3%) received periprocedural positive pressure ventilation during their ED stay. The nature and timing of ven- tilation support are reported in Table 3. Five of the 6 pneumothoraces occurred in patients receiving mechanical ventilation either during the line placement or just after (Table 2). The incidence of pneumothorax was higher among patients receiving positive pressure ventilation than in those who did not (1.6% vs 0.1%, P b .01). Table 3 reports the as- sociation of site of catheterization, initial vein success, and positive pres- sure ventilation with pneumothorax.

All 6 of the pneumothoraces received pleural space drainage: 4 re- ceived anteriorly placed Pigtail catheters with a Heimlich valve, and 2 received laterally placed tube thoracostomies. Outcomes of these drain- age procedures are reported in Table 2. The effect of 1 pigtail catheter on a mechanically ventilated patient was deemed unsuccessful on chest ra- diograph the following day and was replaced by tube thoracostomy.

The number of needle passes was poorly documented, as was the use of static or dynamic ultrasound. These variables were therefore not reported. Percentage of agreement between the 2 abstractors on site of vein access, Catheterization success, and pneumothorax was 100%.

Limitations

This study suffers from the biases and data collection limita- tions that accompany a retrospective design. These are mitigated, however, by the blinding of the abstractors to the study’s secondary

Table 2

Characteristics of ED cases with iatrogenic pneumothorax subsequent to thoracic central venous catheterization

Patient	Attempted initial	Type and timing of	Type of post-CVC	Size (AID)	Degree of	Initial means	Pneumothorax outcome
demographics Age (y) sex	vein access, side (if unsuccessful, second site; timing)	positive pressure ventilation	CXR; side of PTX		mediastinal shift	of ED evacuation
91 M	Subclavian, right	None	Upright; right	25% (2.3)	Small	Pigtail catheter	Pigtail catheter removed on

63 F Subclavian, right (IJ, right; before diagnostic CXR)

80 M IJ, left (IJ, left; after diagnostic CXR)

Noninvasive before and during CVC; invasive after CVC and before CXR

Invasive before, during, and after CVC

hospital day 8 without sequelae Supine; right 34% (2.5) Moderate Pigtail catheter Pigtail catheter removed on

hospital day 7 without sequelae

Supine; left 30% (2.1) Small Tube thoracostomy Unclear: the patient expired the

next day from Refractory septic shock with the chest tube in place

65 F Subclavian, left Invasive before, during, and after CVC

80 F Subclavian, right Invasive before, during,

and after CVC

76 F Subclavian, right Invasive before, during,

and after CVC

Supine; left 39% (3.0) Large Tube thoracostomy Unclear: the patient expired the

next day from refractory septic shock and ischemic bowel with the chest tube in place

Supine; right 50% (4.7) Large Pigtail catheter Pigtail catheter removed on

hospital day 6 without sequelae Supine 29% (2.0) Small Pigtail catheter Pneumothorax persisted despite

pigtail; inpatient tube thoracostomy performed the next day; removed on hospital day 10 without sequelae

CVC, central venous catheterization; IJ, internal jugular; CXR, chest radiograph; AID, average intrapleural distance (cm); PTX, pneumothorax.

aim, the redundancy of our data collection and chart review methods, and our high interrater reliability.

The physician population may limit the study’s generalizability. The proceduralists were predominately residency-trained, board-certified emergency physicians who work in medium and large nonrural commu- nity EDs. The setting, training, and experience of the emergency physi- cians may not reflect those of smaller rural EDs in the US or large academic level I trauma centers [13]. This limitation is tempered by the di- verse facility with central venous catheterization these same physicians report. A survey of this population (78% response rate, 367/472) per- formed in 2010, during the study period, found that 15% stated that they were “not really comfortable placing thoracic central lines.” When asked about their experience in the entire preceding year, 12% reported placing 2 or fewer central lines, whereas 20% reported placing 20 or more [14]. This broad diversity might expand the study’s external validity. Another shortcoming is our inability to report rates of ultrasound use. We know from the aforementioned survey, however, that a large

number of our emergency physicians use adjunct ultrasound to place thoracic central lines [14]. Ultrasound use has been shown to reduce the complication rate of thoracic central venous catheterization [1,2]. A recent large meta-analysis of randomized studies comparing real- time ultrasound guidance with anatomic landmark techniques found that ultrasound significantly reduced the risk of cannulation failure (pooled relative risk, 0.18; 95% CI, 0.10-0.32; P b .001) and reduced the occurrence of iatrogenic pneumothorax (pooled relative risk, 0.21; 95% CI, 0.06-0.73; P = .01) [2].

Lastly, we had too few outcomes to undertake multivariate analysis, preventing us from identifying the independent contribution of our pre- dictor variables to the pneumothorax outcome.

Discussion

This large Multicenter retrospective study found that the overall in- cidence of iatrogenic pneumothorax among community ED patients

Table 3

Association of site of catheterization, initial vein success, and positive pressure ventilation with pneumothorax in ED patients undergoing thoracic central venous catheterization

	Cases N = 1249 n (%)	Immediate pneumothorax N = 6 n	Pneumothorax incidence (%) (95% CI)	P value
Vein of catheterization attempt
Internal jugular only	1032 (82.6)	1	0.1 (0-0.6)	b.001
Subclavian only or subclavian with internal jugular	217 (17.4)	5	2.3 (0.8-5.4)
Success at initial thoracic central vein site Yes	1168 (93.5)	4	0.3 (0.1-0.9)	.05
No	81 (6.5)	2	2.5 (0.2-9.1)

Periprocedural positive pressure ventilation

No 933 (74.7) 1 0.1 (0-0.7) b.01

Yes	316 (25.3) 5a 1.6 (0.6-3.8)
Invasive	234 (74.1)
Before/during catheterization	175
Initiated after catheterizationb	59
Noninvasive	82 (25.9)
Before/after catheterizationc	61
Initiated after catheterization	21

^a Four cases were receiving mechanical ventilation before, during, and after central venous catheterization; 1 case was receiving noninvasive positive pressure ventilation before and during central venous catheterization and then mechanical ventilation after the catheterization (Table 2).

^b Seven of the 59 cases who received invasive positive pressure ventilation after placement of a thoracic central venous catheter had also received precatheterization noninvasive positive pressure ventilation; one of these developed a pneumothorax. Because of incomplete documentation, we cannot report how many of these 7 cases had their noninvasive positive pressure ventilation interrupted during the thoracic central vein catheterization itself.

^c Because of incomplete documentation, we cannot report how many of these 61 cases had their noninvasive positive pressure ventilation interrupted during the thoracic central vein catheterization itself.

undergoing attempted thoracic central venous catheterization was very low. The rate was significantly higher with the subclavian vein com- pared with the internal jugular vein approach, with failed access at the initial vein site, and with periprocedural positive pressure ventilation.

Incidence of immediate and delayed pneumothorax

Ours is one of the first studies to report the incidence of pneumotho- rax complicating thoracic central line placement in adults in the com- munity ED setting. The overall low incidence we observed (0.5%), however, is consistent with the rates reported among adults in academ- ic critical care settings (Table 4).

Seven smaller randomized clinical trials in the last decade assessed pneumothorax risk in adults [10,24-28], as reported in a recent meta- analysis [2]. The combined incidence among 1752 adults was 1.6% (95% CI, 1.1%-2.3%). Small prospective studies in academic EDs demon- strate a similarly low incidence rate of pneumothorax complicating tho- racic central venous catheterization [6,10,11].

The studies described in Table 4 did not report the rate of delayed pneumothorax, that is, cases with an initially negative chest radiograph result that were found on subsequent imaging hours to days later dur- ing their inpatient stay to have a pneumothorax. Some delayed pneumothoraces are detected when the patient develops suggestive symptoms or when they undergo imaging for other purposes and the pneumothorax is found incidentally. In our cohort, we searched the physician discharge summaries and the discharge diagnoses for evi- dence of delayed pneumothorax not attributable to a subsequent post-ED procedure (eg, thoracentesis). We identified only one delayed pneumothorax, detected during an inpatient computerized tomography of the abdomen, 40 hours after the ED central venous catheterization. Occult pneumothoraces are thought to be more common than recog- nized and are attributable to the insensitivity of the postprocedure chest radiograph [20,29]. Because the detection rate of occult pneumothoraces in a delayed fashion is low, these are often published as case reports [30-37]. One large retrospective study of 9637 outpa- tients with cancer receiving subclavian lines for chemotherapy found a low incidence of delayed pneumothorax (0.4%; 95% CI, 0.3%-0.5%) [38]. Patients in this series had been reimaged because of persistent or worsening symptoms or 24 hours after an unsuccessful insertion when a new central line was placed in the opposite subclavian vein.

Internal jugular vs subclavian

The lower rate of pneumothorax we observed in internal jugular vein compared with subclavian vein catheterizations (approximately 0.1% vs 2.3%) is explicable anatomically given the greater proximity of the subcla- vian vein to the pleural dome. Our findings replicate what has been re- ported in recent large studies. Iovino et al [39] found similarly disparate rates of pneumothorax when comparing internal jugular with subclavian

vein catheterizations: 0 of 1113 (0%; 95% CI, 0%-0.4%) vs 17 of 554 (3.1%; 95% CI, 1.9%-4.9%) (P b .0001). Roux et al [40] reported a similar vein- specific pneumothorax rate discrepancy: 0.4% (95% CI, 0.1%-1.0%) of

1065 internal jugular vein catheterizations and 2.3% (95% CI, 1.3%-3.8%) of 621 subclavian vein catheterizations (P b .001). Pikwer et al [41] docu- mented a rate of pneumothorax of 0.3% (95% CI, 0.1%-0.7%) among 1552 internal jugular vein catheterizations and 1.6% (95% CI, 0.7%-3.3%) among 424 subclavian vein catheterizations. Other studies concur: with regard to the risk of pneumothorax, the internal jugular vein approach is sub- stantially safer than the subclavian vein approach [12,42,43].

Failed catheterization at the initial vein

We observed that patients whose first thoracic vein site was unsuc- cessfully cannulated had higher rates of iatrogenic pneumothorax. A failed catheterization attempt has been shown to be the strongest pre- dictor of a line-related complication [44]. Each additional needle pass increases the mechanical complication rate [42,45], which is why some have recommended that the number of needle passes be restrict- ed [46]. Ultrasound use reduces the risk of mechanical complications, pneumothorax included, by improving the odds of first-pass catheteri- zation success [1,2,8,10,25,47].

The effect of positive pressure ventilation

We found a higher incidence of venous catheter-related pneumo- thorax among patients undergoing ED positive pressure ventilation. Positive pressure ventilation contributes to the development of pneu- mothorax in 2 ways. First, barotrauma from positive pressure ventila- tion alone can cause pneumothorax, independent of iatrogenic pleural violation during central venous catheterization [3]. This occurs more commonly in patients with underlying chronic pulmonary disease [48]. Second, positive pressure ventilation can enlarge the size of pneumothoraces, even transforming what would have been occult ab- normalities into pneumothoraces sufficiently large to be detectable by plain chest radiograph [34]. Such a process also has been demonstrated in patients with Traumatic pneumothoraces [20]. Positive pressure ven- tilation also increases the risk for tension pneumothorax, which is why drainage of even small pneumothoraces is customarily recommended in Mechanically ventilated patients [48,49]. A pneumothorax in a me- chanically ventilated patient carries a greater risk of serious barotrauma, which can even be life threatening [50].

Randomized trials undertaken on mechanically ventilated patients have found higher rates of iatrogenic pneumothorax than those report- ed above (Table 4) when experienced providers were using anatomic landmark techniques: 5.0% (10/201) with the subclavian approach

[28] and 2.4% (11/450) with the internal jugular approach [25]. Those rates are significantly lower in studies in which real-time ultrasound guidance was used: 0 of 201 (95% upper confidence limit, 2.3%) with

Table 4

Rate of immediate pneumothorax from thoracic central venous catheterization in large studies of adults (n N 1000)

Author (year) (ref)	Design	Setting	Proceduralists	Cases n	Veins	Ultrasound guidance	PTX n	% (95% CI)
Iovino et al (2001) [39]	Prospective multicenter	Perioperative or	Surgical faculty	1667	IJ and SC	No	17	1.0 (0.6-1.6)
oncology-hematology department
Pikwer et al (2009) [41]	Prospective single center	Mostly operating room,	Anesthesia and intensive	1976a	IJ and SC	No	13	0.6 (0.3-1.0)
		perioperative, or	care faculty and directly
		intensive care unit	supervised residents
Cavanna et al (2010) [58]	Prospective single center	Oncology-hematology	2 oncologists and a nurse	1978	IJ	Yes	0	0 (0-0.2)
		department
Roux et al (2014) [40]	Prospective multicenter	Intensive care unit	Residents of multiple	1686	IJ and SC	+-	18	1.1 (0.7-1.7)

IJ, internal jugular; SC, subclavian; PTX, pneumothorax.

^a 124 patients were less than 18 years old.

specialties, direct supervision in ~ 23%

the subclavian approach and 0 of 450 (95% upper confidence limit, 1.0%) with the internal jugular approach. A prospective case series of 290 me- chanically ventilated patients undergoing subclavian vein catheteriza- tion via the supraclavicular approach found no cases of pneumothorax (95% upper confidence limit, 1.6%) [51].

Interestingly, few studies of pneumothorax complications from cen- tral venous catheterization even report the number of patients who were receiving periprocedural positive pressure ventilation, much less calculate the association of positive pressure ventilation with the inci- dence of iatrogenic pneumothorax. This is true for studies both of inten- sive care patients [27,42,52-54] and ED patients [7,10,11].

The risk of iatrogenic pneumothorax in our cohort was highest among mechanically ventilated patients undergoing central venous catheterization via the subclavian approach. As data from the literature suggest, the safest way to place a thoracic central line in a mechanically ventilated patient is to access the internal jugular vein using real-time ul- trasound guidance to increase the odds of first-stick success [25]. positive end-expiratory pressure should be removed throughout the procedure, and the puncture should be performed during the expiratory phase [52]. In a multicenter study in our health care delivery system, we found that 28% of 593 septic patients with an indication for central venous catheterization refused the procedure [15]. Their refusal to give in- formed consent is thought to be dependent in large part on the physi- cian’s portrayal of the procedure’s risks and benefits. The high rate of refusal suggests that we might be misrepresenting the true risks associ- ated with the procedure. This current study was undertaken to deter- mine a more accurate estimate of the risks that attend thoracic central line placement. Our results comport with the literature to suggest that ED patients undergoing central venous catheterization by the internal jugular approach have a Very low risk (0.1%) for developing a pneumo- thorax. Major bleeding is another mechanical complication of central venous catheterization that is also uncommon. One recent study from our research network demonstrated that the risk of major hemorrhage from central venous catheterization in patients with coagulopathy or thrombocytopenia is also very low (0.1%; 95% upper confidence limit, 0.6%) [16]. Knowing the rates of these mechanical complications can help inform our own risk/benefit calculus when considering thoracic central venous catheterization in the community ED setting. This knowledge can also shape the manner in which we explain the proce-

dure to our patients as part of the informed consent process.

The evacuation of pneumothoraces

All of our patients with a pneumothorax were treated with immediate evacuation, either by pigtail catheter or by tube thoracostomy. Studies suggest that small pneumothoraces (generally b 15%) in asymptomatic, Hemodynamically stable patients may require no treatment at all or at most High-flow oxygen through a nonrebreather mask to facilitate air re- sorption from the pleural space [38]. Patients receiving positive pressure ventilation are an exception, as even a small pneumothorax in a ventilat- ed patient warrants evacuation to avoid enlargement and even transfor- mation into a tension pneumothorax. The traditional method of pneumothorax evacuation had been tube thoracostomy. However, recent evidence suggests that a small pigtail catheter (14F) is just as effective as a larger chest tube (28F) [55] but with significantly less pain [56,57]. Our numbers are too small to comment on the relative effectiveness of the dif- ferent evacuation approaches.

In sum, we found that the incidence of immediate pneumothorax from thoracic central venous catheterization in community EDs is approximately 0.5%, a rate consistent with large prospective studies in academic settings. The risk is higher in patients undergoing a sub- clavian vein approach, in patients whose first vein site was unsuc- cessfully cannulated, and in patients receiving positive pressure ventilation. These data can help direct the risk/benefit calculus and the informed consent process for ED patients with indications for thoracic central venous catheterization.

Acknowledgments

We are grateful to the Kaiser Permanente Northern California Com- munity Benefit Program and the Gordon and Betty Moore Foundation for their financial support. We also thank YunYi Hung, PhD, Kaiser Permanente Division of Research, for help with programming. We are also grateful to Francesco Iovino, MD, PhD, Assistant Professor, Division of General Surgery, Department of Anesthesiological, Surgical and Emergency Sciences, Second University of Naples, Naples, Italy, for pro- viding unpublished details from his research on the safety of central ve- nous catheterization.

References

1. Lamperti M, Bodenham AR, Pittiruti M, Blaivas M, Augoustides JG, Elbarbary M, et al. International evidence-based recommendations on ultrasound-guided vascular access. Intensive Care Med 2012;38:1105-17.
2. Wu SY, Ling Q, Cao LH, Wang J, Xu MX, Zeng WA. Real-time two-dimensional ultra- sound guidance for central venous cannulation: a meta-analysis. Anesthesiology 2013;118:361-75.
3. AHRQ quality indicators. Patient safety indicators #6: technical specifications. Iatrogenic pneumothorax rate [version 4.4]. Rockville (MD): Agency for Healthcare Research and Quality (AHRQ); 2012.
4. Glickman SW, Krubert C, Koppenhaver J, Glickman LT, Schulman KA, Cairns CB. Increased rate of central venous catheterization procedures in community EDs. Am J Emerg Med 2010;28:208-12.
5. Theodoro D, Owens PL, Olsen MA, Fraser V. Rates and timing of central venous can- nulation among patients with sepsis and respiratory arrest admitted by the emer- gency department. Crit Care Med 2014;42:554-64.
6. Hrics P, Wilber S, Blanda MP, Gallo U. Ultrasound-assisted internal jugular vein cath- eterization in the ED. Am J Emerg Med 1998;16:401-3.
7. Miller AH, Roth BA, Mills TJ, Woody JR, Longmoor CE, Foster B. Ultrasound guidance versus the landmark technique for the placement of central venous catheters in the emergency department. Acad Emerg Med 2002;9:800-5.
8. Milling Jr TJ, Rose J, Briggs WM, Birkhahn R, Gaeta TJ, Bove JJ, et al. Randomized, con- trolled clinical trial of point-of-care limited ultrasonography assistance of central ve- nous cannulation: the Third Sonography Outcomes Assessment Program (SOAP-3) Trial. Crit Care Med 2005;33:1764-9.
9. Milling T, Holden C, Melniker L, Briggs WM, Birkhahn R, Gaeta T. Randomized con- trolled trial of single-operator vs. two-operator ultrasound guidance for internal jug- ular central venous cannulation. Acad Emerg Med 2006;13:245-7.
10. Leung J, Duffy M, Finckh A. Real-time ultrasonographically-guided internal jugular vein catheterization in the emergency department increases success rates and re- duces complications: a randomized, prospective study. Ann Emerg Med 2006;48: 540-7.
11. Theodoro D, Bausano B, Lewis L, Evanoff B, Kollef M. A descriptive comparison of ultrasound-guided central venous cannulation of the internal jugular vein to landmark-based subclavian vein cannulation. Acad Emerg Med 2010;17:416-22.
12. Theodoro D, Krauss M, Kollef M, Evanoff B. Risk factors for acute adverse events dur- ing ultrasound-guided central venous cannulation in the emergency department. Acad Emerg Med 2010;17:1055-61.
13. Centers for Disease Control and Prevention. Table 106. Hospitals, beds, and occupancy rates, by type of ownership and size of hospital: United States, selected years 1975-2010. http://www.cdc.gov/nchs/data/hus/2012/106.pdf. [Accessed September 21, 2014].
14. Ballard DW, Reed ME, Rauchwerger AS, Chettipally UK, Offerman SR, Mark DG, et al. Emergency Physician Perspectives on Central Venous Catheterization in the Emer- gency Department: A survey-based study. Acad Emerg Med 2014;21:623-30.
15. Vinson DR, Ballard DW, Stevenson MD, Mark DG, Reed ME, Rauchwerger AS, et al. Predictors of Unattempted Central Venous Catheterization in Septic Patients Eligible for Early Goal-directed Therapy. West J Emerg Med 2014;15:67-75.
16. Vinson DR, Ballard DW, Hance LG, Hung Y, Rauchwerger AS, Reed ME, et al. Bleeding complications of central venous catheterization in septic patients with abnormal hemostasis. Am J Emerg Med 2014;32:737-42.
17. Bornstein S. An integrated EHR at Northern California Kaiser Permanente: Pitfalls, challenges, and benefits experienced in transitioning. Appl Clin Inform 2012;3: 318-25.
18. Whippy A, Skeath M, Crawford B, Adams C, Marelich G, Alamshahi M, et al. Kaiser Permanente’s performance improvement system, part 3: multisite improvements in care for patients with sepsis. Jt Comm J Qual Patient Saf 2011;37:483-93.
19. Graham AS, Ozment C, Tegtmeyer K, Lai S, Braner DA. Videos in clinical medicine. Central venous catheterization. N Engl J Med 2007;356:e21.
20. Ball CG, Kirkpatrick AW, Feliciano DV. The Occult pneumothorax: what have we learned? Can J Surg 2009;52:E173-9.
21. Tukey MH, Borzecki AM, Wiener RS. Validity of ICD-9-CM Codes for the Identifica- tion of Complications Related to Central Venous Catheterization. Am J Med Qual 2013. http://dx.doi.org/10.1177/1062860613512518 [Epub ahead of print].
22. Choi BG, Park SH, Yun EH, Chae KO, Shinn KS. Pneumothorax size: correlation of su- pine anteroposterior with erect posteroanterior chest radiographs. Radiology 1998; 209:567-9.
23. Collins CD, Lopez A, Mathie A, Wood V, Jackson JE, Roddie ME. Quantification of pneumothorax size on chest radiographs using interpleural distances: regression

    analysis based on volume measurements from helical CT. AJR Am J Roentgenol 1995;165:1127-30.

    Cajozzo M, Quintini G, Cocchiera G, Greco G, Vaglica R, Pezzano G, et al. Comparison of central venous catheterization with and without ultrasound guide. Transfus Apher Sci 2004;31:199-202.

24. Karakitsos D, Labropoulos N, De Groot E, Patrianakos AP, Kouraklis G, Poularas J, et al. Real-time ultrasound-guided catheterisation of the internal jugular vein: a prospec- tive comparison with the landmark technique in critical care patients. Crit Care 2006;10:R162.
25. Agarwal A, Singh DK, Singh AP. Ultrasonography: a novel approach to central venous cannulation. Indian J Crit Care Med 2009;13:213-6.
26. Palepu GB, Deven J, Subrahmanyam M, Mohan S. Impact of ultrasonography on central venous catheter insertion in intensive care. Indian J Radiol Imaging 2009;19:191-8.
27. Fragou M, Gravvanis A, Dimitriou V, Papalois A, Kouraklis G, Karabinis A, et al. Real-time ultrasound-guided subclavian vein cannulation versus the landmark method in critical care patients: a prospective randomized study. Crit Care Med 2011;39:1607-12.
28. Ghane MR, Saburi A, Javadzadeh HR. A recommended method in order to interpret chest X-rays for diagnosing small size pneumothorax. Int J Crit Illn Inj Sci 2013;3:36-9.
29. Mitchell A, Steer HW. Late appearance of pneumothorax after subclavian vein cath- eterisation: an anaesthetic hazard. Br Med J 1980;281:1339.
30. Slezak FA, Williams GB. Delayed pneumothorax: a complication of subclavian vein catheterization. JPEN J Parenter Enteral Nutr 1984;8:571-4.
31. Sivak SL. Late appearance of pneumothorax after subclavian venipuncture. Am J Med

    1986;80:323-4.

    Plaus WJ. Delayed pneumothorax after subclavian vein catheterization. JPEN J Parenter Enteral Nutr 1990;14:414-5.

32. Cronen MC, Cronen PW, Arino P, Ellis K. Delayed pneumothorax after subclavian vein catheterization and positive pressure ventilation. Br J Anaesth 1991;67:480-2.
33. Spiliotis J, Kordossis T, Kalfarentzos F. The incidence of delayed pneumothorax as a complication of subclavian vein catheterisation. Br J Clin Pract 1992;46:171-2.
34. Tyburski JG, Joseph AL, Thomas GA, Saxe JM, Lucas CE. Delayed pneumothorax after central venous access: a potential hazard. Am Surg 1993;59:587-9.
35. Plewa MC, Ledrick D, Sferra JJ. Delayed tension pneumothorax complicating central venous catheterization and positive pressure ventilation. Am J Emerg Med 1995;13: 532-5.
36. Laronga C, Meric F, Truong MT, Mayfield C, Mansfield P. A treatment algorithm for pneumothoraces complicating central venous catheter insertion. Am J Surg 2000; 180:523-6 [discussion 6-7].
37. Iovino F, Pittiruti M, Buononato M, Lo Schiavo F. Central venous catheterization: complications of different placements. Ann Chir 2001;126:1001-6.
38. Roux D, Reignier J, Thiery G, Boyer A, Hayon J, Souweine B, et al. Acquiring procedur- al skills in ICUs: a prospective multicenter study. Crit Care Med 2014;42:886-95.
39. Pikwer A, Baath L, Perstoft I, Davidson B, Akeson J. Routine chest X-ray is not re- quired after a low-risk central venous cannulation. Acta Anaesthesiol Scand 2009; 53:1145-52.
40. Eisen LA, Narasimhan M, Berger JS, Mayo PH, Rosen MJ, Schneider RF. Mechanical complications of central venous catheters. J Intensive Care Med 2006;21:40-6.
41. Fisher NC, Mutimer DJ. Central venous cannulation in patients with liver disease and coagulopathy-a prospective audit. Intensive Care Med 1999;25:481-5.
42. Mansfield PF, Hohn DC, Fornage BD, Gregurich MA, Ota DM. Complications and fail- ures of subclavian-vein catheterization. N Engl J Med 1994;331:1735-8.
43. McGee DC, Gould MK. Preventing complications of central venous catheterization. N Engl J Med 2003;348:1123-33.
44. Takeyama H, Taniguchi M, Sawai H, Funahashi H, Akamo Y, Suzuki S, et al. Limiting vein puncture to three needle passes in subclavian vein catheterization by the infraclavicular approach. Surg Today 2006;36:779-82.
45. Hind D, Calvert N, McWilliams R, Davidson A, Paisley S, Beverley C, et al. Ultrasonic locating devices for central venous cannulation: meta-analysis. BMJ 2003;327:361.
46. Baumann MH, Noppen M. Pneumothorax. Respirology 2004;9:157-64.
47. Giacomini M, Iapichino G, Armani S, Cozzolino M, Brancaccio D, Gallieni M. How to avoid and manage a pneumothorax. J Vasc Access 2006;7:7-14.
48. Esteban A, Anzueto A, Frutos F, Alia I, Brochard L, Stewart TE, et al. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day interna- tional study. JAMA 2002;287:345-55.
49. Czarnik T, Gawda R, Perkowski T, Weron R. Supraclavicular approach is an easy and safe method of subclavian vein catheterization even in mechanically ventilated pa- tients: analysis of 370 attempts. Anesthesiology 2009;111:334-9.
50. Lefrant JY, Muller L, De La Coussaye JE, Prudhomme M, Ripart J, Gouzes C, et al. Risk factors of failure and immediate complication of subclavian vein catheterization in critically ill patients. Intensive Care Med 2002;28:1036-41.
51. Merrer J, De Jonghe B, Golliot F, Lefrant JY, Raffy B, Barre E, et al. Complications of femoral and subclavian venous catheterization in critically ill patients: a randomized controlled trial. JAMA 2001;286:700-7.
52. Sznajder JI, Zveibil FR, Bitterman H, Weiner P, Bursztein S. Central vein catheteriza- tion. Failure and complication rates by three percutaneous approaches. Arch Intern Med 1986;146:259-61.
53. Kulvatunyou N, Vijayasekaran A, Hansen A, Wynne JL, O’Keeffe T, Friese RS, et al. Two-year experience of using pigtail catheters to treat traumatic pneumothorax: a changing trend. J Trauma 2011;71:1104-7.
54. Kulvatunyou N, Erickson L, Vijayasekaran A, Gries L, Joseph B, Friese RF, et al. Ran- domized clinical trial of pigtail catheter versus chest tube in injured patients with uncomplicated traumatic pneumothorax. Br J Surg 2014;101:17-22.
55. Voisin F, Sohier L, Rochas Y, Kerjouan M, Ricordel C, Belleguic C, et al. Ambulatory management of large Spontaneous pneumothorax with pigtail catheters. Ann Emerg Med 2014;64:222-8.
56. Cavanna L, Civardi G, Vallisa D, Di Nunzio C, Cappucciati L, Berte R, et al. Ultrasound- guided central venous catheterization in cancer patients improves the success rate of cannulation and reduces mechanical complications: a prospective ob- servational study of 1,978 consecutive catheterizations. World J Surg Oncol 2010;8:91.