a b s t r a c t

Objective: The objective was to determine risk factors associated with difficult venous access (DVA) in the emergency department (ED).

Methods: This was a prospective, observational study conducted in the ED of an urban tertiary care hospital. Adult patients undergoing intravenous (IV) placement were consecutively enrolled during periods of block enrollment. The primary outcome was DVA, defined as 3 or more IV attempts or use of a method of rescue vascular access to establish IV access. univariate and multivariate analyses for factors predicting DVA were performed using logistic regression.

Results: A total of 743 patients were enrolled, of which 88 (11.8%) met the criteria for DVA. In the adjusted analysis, only 3 medical conditions were significantly associated with DVA: diabetes (odds ratio [OR] 1.72, 95% confidence interval [CI] 1.1-2.8), Sickle cell disease (OR 3.8, 95% CI 1.5-9.5), and history of IV drug abuse (OR 2.5, 95% CI 1.1-5.7). Notably, age, body mass index, and dialysis were not. Of patients who reported a history of requiring multiple IV attempts in the past for IV access, 14% met criteria for DVA on this visit (OR 7.7 95% CI 3-18). Of the patients who reported a history of IV insertion into the external jugular, ultrasound-guided IV placement, or a central venous catheter for IV access, 26% had DVA on this visit (OR 16.7, 95% CI 6.8-41).

Conclusions: Nearly 1 of every 9 to 10 adults in an urban ED had DVA. Diabetes, IV drug abuse, and sickle cell disease were found to be significantly associated with DVA.

(C) 2014

Introduction

Background and importance

Intravenous access is the most commonly performed procedure in the emergency department (ED). According to the 2010 National Hospital Ambulatory Care Survey, more than one- quarter (35.2 million) of all ED visits in the United States results in an IV placement for parenteral fluid administration [1]. Rapidly estab- lishing an IV catheter can be challenging, particularly in the subset of patients with a lack of readily visible or palpable veins, a condition known as difficult venous access (DVA). Difficult venous access has been most often defined as at least 2 failed IV attempts [2-9]. Alternative or “rescue” methods of establishing vascular access such as accessing veins in atypical locations (most commonly, the external jugular or “EJ” vein), ultrasound-guided (USG) IV placement, or use of an intraosseous (IO) device are frequently used in DVA patients. In worst case scenarios, central venous catheters are needed [10,11].

The causes of DVA in adult patients are not well understood. A number of studies have reported characteristics believed to contribute to DVA in their sample including IV drug abuse (IVDA), dialysis, sickle

? Previously presented as a Poster Presentation at the Society of Academic Emergency Medicine Scientific Assembly, May 2013, Atlanta, GA.

* Corresponding author. 1020 Sansom St, Thompson Bldg, Room 239, Philadelphia, PA 19107.

cell disease (SCD), chemotherapy, and other chronic medical conditions [2-6,9,10,12-17]. Unfortunately, in these studies, there is no comparison to a non-DVA cohort, limiting the conclusion that these conditions are truly associated with DVA. Two studies from France have examined the association of body mass index (BMI) and first attempt success, but had conflicting results [18,19]. Understand- ing risk factors for DVA may lead to a better understanding of this condition and improve care in this population.

Goals of this investigation

The current study set out to determine risk factors for DVA in adult patients presenting to the ED.

Methods

Study design and setting

This was a prospective observational study conducted from July 2012 to February 2013. The institutional review board of the study institution granted study approval. The study was conducted in the ED of an urban academic hospital with annual census of 65,000 and an emergency medicine residency program. The usual practice patterns for obtaining venous access in our department are as follows: (1) ED staff determines need for IV access; (2) ED technician or nurse performs IV insertion; and (3) in cases of repeated IV placement

http://dx.doi.org/10.1016/j.ajem.2014.07.008

0735-6757/(C) 2014

1180 J.M. Fields et al. / American Journal of Emergency Medicine 32 (2014) 1179-1182

failure or perceived DVA by staff, an emergency medicine resident or attending physician performs rescue vascular access with an USG or EJ

IV. In rare cases, IO and CVCs are used to obtain venous access.

Selection of participants

This study included adult patients presenting to the ED who required IV placement. Consecutive patients who had an IV placed or attempted during their ED stay were enrolled during periods of block enrollment. All patients underwent informed verbal consent. Poten- tially critically ill patients and patients who refused IV placement or data collection were excluded from the study. Trained research associates went to each room in the ED sequentially and consecutively to determine when a patient had an IV placed or attempted. After an IV was placed, the patient was then consented and enrolled.

Methods and measurements

Patient variables potentially associated with DVA were identified based on review of the literature: age, sex, race, BMI, history of chemotherapy, diabetes, dialysis, IVDA, swelling, SCD, and ED visit or hospitalization within the last 90 days [2-10,12-16,18,19]. To analyze the effect of these 12 independent variables in a logistic regression model, the required sample size was 705 patients, which was calculated using the technique by Peduzzi et al [20], assuming a 16% rate of DVA. Goal enrollment was set at 767 patients to allow for the possibility of incomplete data or exclusion. Data on the number of IV attempts as well as the initial size catheter attempted were collected and verified by the patient’s nurse.

Outcomes

The primary outcome variable was the presence of DVA defined as requiring 3 or more IV attempts or if rescue vascular access was used at any time. Rescue methods were considered as any nontraditional vascular access technique such as IV placement in the EJ or other non- upper extremity location, USGIV placement, IO device placement, or insertion of a CVC solely for IV access. An IV attempt was defined as the needle piercing the skin. Successful IV placement was an IV attempt that resulted in the ability to flush saline without development of pain or swelling at the insertion site and the ability to use the IV for medical care.

Analysis

patients for analysis. The mean age was 49 years (SD +- 17 years). Four hundred forty-five patients (61%) were female, 352 (50%) were Caucasian, 295 (42%) were African American, and 677 (93%) reported a high school education or greater.

The initial attempt at IV access was successful in 75.6% (562/743) of patients. Of the 181 remaining patients, 6 underwent immediate rescue access with USGIV placement. The second traditional IV attempt was successful in 53% (93/175), leaving 82 patients who met criteria for DVA (2 failed traditional IV attempts). These 82 patients combined with the 6 patients who underwent immediate rescue IV placement after 1 failed IV attempt yielded a total of 88 of 743 patients who met the definition of DVA, resulting in a prevalence of 11.8% (95% CI 9.6%-14.4%). Overall, 22% (19/88) of patients with DVA required rescue vascular access with USGIV placement. No patients in our sample underwent EJ, IO, or CVC.

A univariate analysis on all variables collected was performed between patients with and without DVA (Table 1). A multivariate logistic regression including demographics and preexisting medical conditions identified 3 independent risk factors for DVA: diabetes, IVDA, and SCD (Table 2). A history of DVA increased the likelihood of DVA on this visit, but depended on the definition. Of patients who reported a history of requiring “multiple IV attempts” in the past for IV access, 14% met criteria for DVA on this visit (OR 7.7, 95% CI 3.0-18.0). Of patients who reported previously requiring an EJ, USGIV, or CVC for IV access, 26% met criteria for DVA (OR 16.7, 95% CI 6.8-41.0). In addition, use of a smaller IV catheter (22 or 24 gauge) for the first attempt was significantly associated with DVA (OR 6.4, 95% CI 3.4-11.9).

4. Discussion

Difficult venous access is a common problem in the ED that can be frustrating for both patients and providers. Little is known about which demographic variables or preexisting conditions are associated with it. The current study is the first to compare the frequency of conditions in a DVA and non-DVA cohort. The prevalence of DVA in the study population was 11.8%, which is similar to previous studies reporting a range of 8% to 16% [16,18]. Of the 12 variables examined in this study, only diabetes, IVDA, and SCD were significant in a multivariate logistic regression model for predicting DVA. Previous factors thought to be associated with DVA-age, history of dialysis,

Table 1

Patient demographics and prevalence of DVA

catheter size initially chosen for traditional IV placement was dichotomized into “large” (20 gauge or larger) and “small” (22 gauge or smaller). A history of DVA was divided into 3 logical categories: none, history of multiple IV attempts, or history of multiple IV attempts with the need for a method of rescue vascular access. The association between each variable and the main outcome

Univariate analysis of the collected variables was performed using	Variable	All		+DVA		-DVA	Diff	P value
?2 and Fisher exact test where appropriate. For simplicity, the IV		N		n (%)		n (%)

All Age, y	743	88 (11.8)	655 (88.2)
18-34	165	23 (13.9)	142 (86.1)	+2.1%	.82
35-54	284	32 (11.3)	252 (88.7)	-0.5%
55-64	148	16 (10.8)	132 (89.2)	-1.0%
>= 65 Sex	146	17 (11.6)	129 (88.4)	-0.2%

(presence of DVA) is reported with frequencies, 95% confidence	Male	294	35 (11.9)	259 (88.0)	+0.1%	.94
intervals (CIs), and P values.	Female	449	53 (11.8)	396 (88.2)	-
Multivariate logistic regression was performed on the 12 patient	Race

demographic and medical history variables (history of DVA and initial IV catheter size were not included in this analysis because the goal was to identify demographic risk factors for DVA). The adjusted odds ratios (ORs) of significant variables are reported with 95% CIs.

African American	298	41 (13.8)	257 (86.2)	+2.0%	.35
White	353	39 (11.0)	315 (89.0)	-0.8%
Other Medical conditions	91	8 (8.8)	83 (91.2)	-3.0%
Chemotherapy	76	11 (14.5)	65 (85.5)	+2.7%	.46
Diabetes	171	29 (17.0)	142 (83.0)	+5.2%	.02
Dialysis	22	2 (9.1)	20 (90.9)	-2.7%	.73
IVDA	35	9 (25.7)	26 (74.3)	+13.9%	b.01
Obesity (BMI N 30)	274	39 (14.2)	235 (85.8)	+2.4%	.12
SCD	23	8 (34.8)	15 (65.2)	+23.0%	b.01
Upper extremity swelling	67	15 (22.4)	52 (77.6)	+10.6%	b.01
ED visit in past 90 d	282	39 (13.8)	243 (86.2)	+2.0%	.21
Admission in past 90 d	214	35 (16.4)	179 (83.6)	+4.6%	.02

Results

Characteristics of study subjects

Of 767 patients meeting inclusion criteria, 20 were excluded because of refusal and 4 because of incomplete data, leaving 743

J.M. Fields et al. / American Journal of Emergency Medicine 32 (2014) 1179-1182 1181

Table 2

Adjusted ORs of significant factors associated with DVA

Condition	Adjusted OR	95% CI
Diabetes	2.1	1.3-3.4
IVDA	2.4	1.1-5.3
SCD	3.5	1.4-8.4

chemotherapy, obesity, multiple ED visits, and admissions-were not significant risk factors for DVA in the present study. The findings suggest that the development of DVA may be complex and cannot likely be explained by concomitant conditions alone.

One of the most commonly believed reasons for development of DVA is repetitive vascular trauma. Repetitive needlesticks (due to either medical reasons or IVDA) lead to scarring and stenosis, which may be the explanation for why IVDA and SCD lead to DVA. The association with diabetes and DVA is less well understood but may be explained by its association with multiple other conditions and increased morbidity causing the patient to require frequent medical attention and vascular access. It may be additionally possible that diabetes has a direct effect on veins themselves. One study by Riches et al [21] has demonstrated that saphenous vein grafts have much poorer outcomes in diabetic vs nonDiabetic patients. Vein grafts from diabetic patients demonstrated changes in morphology and lower activity of a GTPase called RhoA[21]. Further research on how diabetes affects peripheral veins on a molecular level may shed light on development of DVA.

One issue with the repetitive vascular access theory of DVA is that other conditions typically associated with repetitive vascular access such as dialysis, chemotherapy, and frequent admissions were not significant in our study. One explanation for this observation could be that the patients in our sample with these conditions may have not undergone as many vascular attempts during their lifetime as those with the other conditions. Determining the actual number of venous access attempts one has had is difficult and prone to recall bias but, if achievable, may be more predictive than the conditions themselves. Furthermore, there may be some patients who have a predisposition to DVA. In the current study, approximately 13% of patients with DVA had none of the listed conditions from Table 1. Anecdotal observations have described some patients with small, fragile, and mobile veins that are difficult to cannulate in the absence of repetitive vascular trauma. Another theoretical risk factor for DVA is a high proportion of subcutaneous tissue in the region of venous cannulation. Using this theory, one would expect that obesity might lead to DVA given the associated increase in subcutaneous tissue. The relationship of obesity and first attempt IV failure has been examined in 2 previous studies with conflicting results. Sebbane et al [19] found that a BMI of either less than 18.5 kg/m² or greater than 30 kg/m² was associated with initial IV failure, whereas Lapostolle et al [18] found no association with BMI and IV failure. The current study also did not find an association between obesity and DVA. A priori, we dichotomized patient’s BMI into obese or nonobese for analysis; but even in a post hoc analysis using the 4 categories by Sebbane et al, we were not able to reproduce the finding that a BMI of less than 18.5 kg/m² or greater than 30 kg/m² results in DVA. The overall weight of the data suggests that obesity is not a risk factor for DVA. The reason obesity has been thought to be related to DVA is due to increased subcutaneous tissue making veins harder to locate and cannulate in the arms [19]. It may be that BMI alone may not be an accurate descriptor of this phenomenon. The fact that there was a trend in our data for patients with a history of swelling involving the upper extremities to have DVA suggests that there may be a relation between the concentration and distribution of subcutaneous tissue on the patient’s arms and presence of DVA. Whether a direct assessment of the percentage body fat of the upper extremities can predict DVA requires

further study.

Despite its possible causes, the ability to predict DVA is best done when more simply assessing whether the patient has some history of DVA (either multiple IV attempts or requiring a rescue method of IV access) and when the physical examination reveals a lack of readily accessible veins. Assessment of venous “favorability” based on visibility and palpability by the operator has been used as a predictor in other studies [19,22]. However, in the study by Yen et al [22], there was poor interrater reliability on the visibility, palpability, and overall difficulty of veins in children. In the study by Sebbane et al [19], there was higher agreement on whether veins had “favorable,” “unfavorable,” or “very unfavorable” veins; however, this was done on 100 separate patients who were not part of the actual study, limiting its applicability. Alternatively, Lapostolle et al [18] showed that when operators used smaller gauge IV catheters (22 gauge or smaller), they were often unsuccessful because a smaller IV was chosen based on the finding of smaller, more difficult veins on physical examination. This observation was also found in our study, with 22- to 24-gauge catheters being associated with a higher failure rate.

4.1. Limitations

Only stable patients were included in the study, and the results are not applicable to critically ill patients. Selection bias within this cohort may have occurred because of patient refusal to participate. Although refusal to participate would likely be random, it is conceivable that patients with DVA may be more likely to refuse if they are upset at having to have multiple IV sticks and delays in receiving medication, suggesting that DVA may be underrepresented in our sample. In addition, although patients were enrolled consecutively during periods of block enrollment, the periods of enrollment tended to occur during weekdays and daytime hours, possibly resulting in some additional selection bias.

Two potentially important variables not assessed in our study were presence of dehydration and the operator’s Experience level. Dehydration has previously been associated with DVA anecdotally because of flattening of veins [5,8,14]. Because of the lack of an objective and reliable method for noninvasively diagnosing dehydra- tion, the current study did not assess this possible predictor in our cohort. Incidentally, patients with Severe dehydration also likely fall into the category of critically ill patients, who were excluded; and thus, the results of our cohort would unlikely be affected. In addition, the current study did not incorporate the level of experience of the operator placing the IV. Theoretically, an inexperienced operator may require more attempts at successful IV cannulation, thereby mis- classifying a patient as having DVA. However, a previous study suggests that this is not the case and did not find an association between operator years of experience and frequency of DVA [18].

5. Conclusion

Difficult venous access is present in approximately 1 out of every 9 to 10 people undergoing IV access in an Urban academic ED. Significant risk factors include diabetes, IVDA, and SCD. Several factors previously thought to be associated with DVA-obesity, dialysis, and chemotherapy-were not found to be significant in the current study.

References

1. National Hospital Ambulatory Medical Care Survey: 2010 Emergency Department Summary Tables. Cent Dis Control 2010 [Available at: http://www.cdc.gov/nchs/ data/ahcd/nhamcs_emergency/2010_ed_web_tables.pdf. Accessed March 20, 2013].
2. Keyes LE, Frazee BW, Snoey ER, Simon BC, Christy D. Ultrasound-guided brachial and basilic vein cannulation in emergency department patients with difficult intravenous access. Ann Emerg Med 1999;34(6):711-4.
3. Mills CN, Liebmann O, Stone MB, Frazee BW. Ultrasonographically guided insertion of a 15-cm catheter into the deep brachial or basilic vein in patients

   1182 J.M. Fields et al. / American Journal of Emergency Medicine 32 (2014) 1179-1182

   with difficult intravenous access. Ann Emerg Med 2007;50(1):68-72. http://dx. doi.org/10.1016/j.annemergmed.2007.02.003.

   Resnick JR, Cydulka RK, Donato J, Jones RA, Werner SL. Success of ultrasound- guided peripheral intravenous access with skin marking. Acad Emerg Med 2008; 15(8):723-30. http://dx.doi.org/10.1111/j.1553-2712.2008.00174.x.

4. Bauman M, Braude D, Crandall C. Ultrasound-guidance vs. standard technique in difficult vascular access patients by ED technicians. Am J Emerg Med 2009;27(2): 135-40. http://dx.doi.org/10.1016/j.ajem.2008.02.005.
5. Panebianco NL, Fredette JM, Szyld D, Sagalyn EB, Pines JM, Dean AJ. What you see (sonographically) is what you get: vein and patient characteristics associated with successful ultrasound-guided peripheral intravenous placement in patients with difficult access. Acad Emerg Med 2009;16(12):1298-303. http://dx.doi.org/10. 1111/j.1553-2712.2009.00520.x.
6. Fields JM, Dean AJ, Todman RW, et al. The effect of vessel depth, diameter, and location on ultrasound-guided peripheral intravenous catheter longevity. Am J Emerg Med 2012;30(7):1134-40. http://dx.doi.org/10.1016/j.ajem.2011.07.027.
7. Schoenfeld E, Boniface K, Shokoohi H. ED technicians can successfully place ultrasound-guided intravenous catheters in patients with poor vascular access. Am J Emerg Med 2011;29(5):496-501. http://dx.doi.org/10.1016/j.ajem.2009.11.021.
8. Au AK, Rotte MJ, Grzybowski RJ, Ku BS, Fields JM. Decrease in Central venous catheter placement due to use of ultrasound guidance for peripheral intravenous catheters. Am J Emerg Med 2012;30(9):1950-4. http://dx.doi.org/10.1016/j.ajem.2012.04.016.
9. Chinnock B, Thornton S, Hendey GW. Predictors of success in nurse-performed ultrasound-guided cannulation. J Emerg Med 2007;33(4):401-5. http://dx.doi. org/10.1016/j.jemermed.2007.02.027.
10. Costantino TG, Parikh AK, Satz WA, Fojtik JP. Ultrasonography-guided peripheral intravenous access versus traditional approaches in patients with difficult intravenous access. Ann Emerg Med 2005;46(5):456-61. http://dx.doi.org/10. 1016/j.annemergmed.2004.12.026.
11. Brannam L, Blaivas M, Lyon M, Flake M. emergency nurses‘ utilization of ultrasound guidance for placement of peripheral intravenous lines in difficult- access patients. Acad Emerg Med 2004;11(12):1361-3. http://dx.doi.org/10.1197/ j.aem.2004.08.027.
12. Stein J, George B, River G, Hebig A, McDermott D. Ultrasonographically guided peripheral intravenous cannulation in emergency department patients with difficult intravenous access: a randomized trial. Ann Emerg Med 2009;54(1): 33-40. http://dx.doi.org/10.1016/j.annemergmed.2008.07.048.
13. Costantino TG, Kirtz JF, Satz WA. Ultrasound-guided Peripheral venous access vs. the external jugular vein as the initial approach to the patient with difficult vascular access. J Emerg Med 2010;39(4):462-7. http://dx.doi.org/10.1016/j. jemermed.2009.02.004.
14. Dargin JM, Rebholz CM, Lowenstein RA, Mitchell PM, Feldman JA. Ultrasonogra- phy-guided peripheral intravenous catheter survival in ED patients with difficult access. Am J Emerg Med 2010;28(1):1-7. http://dx.doi.org/10.1016/j.ajem.2008. 09.001.
15. Elia F, Ferrari G, Molino P, et al. Standard-length catheters vs long catheters in ultrasound-guided peripheral vein cannulation. Am J Emerg Med 2012;30(5):712-6.
16. Schoenfeld E, Shokoohi H, Boniface K. ultrasound-guided peripheral intravenous access in the emergency department: patient-centered survey. West J Emerg Med 2011;12(4):475-7. http://dx.doi.org/10.5811/westjem.2011.3.1920.
17. Lapostolle F, Catineau J, Garrigue B, et al. Prospective evaluation of peripheral venous access difficulty in emergency care. Intensive Care Med 2007;33(8): 1452-7. http://dx.doi.org/10.1007/s00134-007-0634-y.
18. Sebbane M, Claret P-GP-G, Lefebvre S, et al. Predicting peripheral venous access difficulty in the emergency department using body mass index and a clinical evaluation of venous accessibility. J Emerg Med 2013;44(2):299-305. http://dx. doi.org/10.1016/j.jemermed.2012.07.051.
19. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49(12):1373-9.
20. Riches K, Warburton P, O’Regan DJ, Turner NA, Porter KE. Type 2 diabetes impairs venous, but not arterial smooth muscle cell function: possible role of differential RhoA activity. Cardiovasc Revasc Med 2014;15(3):141-8.
21. Yen K, Riegert A, Gorelick MH. Derivation of the DIVA score: a clinical prediction rule for the identification of children with difficult intravenous access. Pediatr Emerg Care 2008;24(3):143-7. http://dx.doi.org/10.1097/PEC.0b013e3181666f32.