The Michigan Appropriateness Guide for Intravenous Catheters (MAGIC): Results From a Multispecialty Panel Using the RAND/UCLA Appropriateness Method

Reliable venous access is a cornerstone of safe and effective care of hospitalized patients. Spurred by technological advances, several venous access devices (VADs) for use during and beyond hospitalization are available to meet this need. In recent years, peripherally inserted central catheters (PICCs) have become popular for venous access in hospital settings (1, 2). Compared with traditional central venous catheters (CVCs), PICCs offer several advantages, including safer insertion in the arm, cost-effective and convenient placement via vascular access nursing teams, and self-care compatibility that facilitates use beyond hospitalization (3–5). It is therefore not surprising that use of PICCs has grown considerably worldwide (6–8).

Despite these advantages, PICCs are central venous catheters that may lead to important complications (9). For instance, problems such as luminal occlusion, malpositioning, and dislodgement occur frequently with these devices (10–12). Similarly, superficial thrombophlebitis or infection at the site of PICC insertion may occur despite uneventful and optimal placement (13, 14). In addition, PICCs are associated with morbid complications, including venous thromboembolism and central line–associated bloodstream infection (15–17). Ensuring appropriate use of PICCs is thus vital to preventing these costly and potentially fatal adverse events.

A growing number of studies suggest substantial variation and potentially inappropriate use of PICCs in hospitalized patients. For example, in a study from a large academic medical center, many PICCs were not actively used or were inserted in patients who also had peripheral intravenous catheters (18). In a decade-long study conducted in a tertiary hospital, changes in patterns of PICC use, including shorter dwell times and ambiguous indications for insertion, were reported (19). Additional cause for concern comes from a recent study, which found that 1 in 5 inpatient providers did not know that their patients had CVCs, with lack of awareness being greatest for PICCs (20). Surveys of inpatient providers have also demonstrated knowledge gaps related to appropriate indications and care practices for PICCs (21, 22). Collectively, these data have not only led to reviews of PICC use in hospitals (23) but also to calls by the Choosing Wisely initiative to improve PICC practices across the United States (24, 25).

The concepts of inappropriate overuse and underuse of medical devices are by no means unique to PICCs. Rather, such issues accompany the diffusion of many novel health technologies. In many such instances, a key barrier to achieving appropriate use is the fact that evidence at a level of detail needed to apply to the range of patients seen in everyday practice is not available. Nevertheless, clinicians must make choices regarding such innovations on a daily basis, potentially fueling inconsistent practice. In the absence of high-quality evidence, an approach that combines available data with the experience and insight of clinical experts is valuable as it would provide guidance where none is otherwise available.

Given this background, we organized and conducted a multidisciplinary meeting of national and international experts to develop appropriateness criteria for use, care, and management of PICCs and related VADs in hospitalized patients. Our objectives were to 1) develop a list of appropriate indications for use of PICCs in relation to other VADs, 2) define the appropriateness of practices associated with the insertion and care of PICCs, 3) determine appropriate practices for treatment and prevention of PICC complications, and 4) rate the appropriateness of peripheral intravenous catheter use in situations that prompt PICC placement.

Methods

Overview of the RAND/UCLA Appropriateness Method

We used the RAND Corporation/University of California Los Angeles (RAND/UCLA) Appropriateness Method to create criteria for appropriate use of PICCs and related VADs (10). Introduced in the 1980s, the RAND/UCLA method was developed to enable measurement of overuse of medical and surgical procedures. According to this methodology, a procedure is considered appropriate when the "expected health benefits (e.g., increased life expectancy, relief of pain, reduction of anxiety or pain) exceed the expected negative consequences (e.g., mortality, morbidity, anxiety, pain) by a sufficiently wide margin such that the procedure is worth doing, exclusive of cost." The approach has thus been applied to an array of procedures, including coronary angiography (26), surgical procedures (27, 28), cataract removal (29), and transplant organ allocation (30). Recently, the method was also used to develop criteria for appropriate use of urinary catheters in hospitalized patients (31).

The RAND/UCLA method was particularly valuable for developing PICC appropriateness criteria for several reasons. First, the approach allowed the synthesis of the best available evidence with practice-based, domain-specific insights from experts. This unique combination ensured both clinical relevance and evidentiary support for the developed recommendations. Second, unlike other group-rating methods, the focus of the RAND/UCLA approach is not to ensure consensus, but minimize artifactual disagreement that may arise from misunderstanding of scenarios being rated. This nuance is highly relevant in the case of PICCs, because available evidence is derived from heterogeneous study designs (for example, retrospective, case–control studies and randomized trials), populations (for example, critically ill, cancer), and clinical specialties (nursing, radiology, medical or surgical disciplines) and is thus prone to misinterpretation. Because the RAND/UCLA method pairs clear instructions and precise clinical definitions with a systematic, reliable, and reproducible rating system (27), the recommendations generated will have high internal validity. Finally, should clinical scenarios lack sufficient detail to make an informed judgment regarding appropriateness, the RAND/UCLA method encourages clarification by panelists so as to make ratings more relevant and precise. In this fashion, generalizability and external validity of the developed appropriateness indications are also ensured.

Proper conduct of the RAND/UCLA Appropriateness Method requires the sequential performance of several steps, including information synthesis, panelist selection, creation of scenarios, rating process, and analysis of results.

Information Synthesis

The first step of the RAND/UCLA Appropriateness Method is to systematically review and synthesize the available literature. With the assistance of 2 research librarians, we searched for English-language articles (between 12 November 2012 and 1 July 2013) by using the following databases: MEDLINE via Ovid (1950 to present), EMBASE (1946 to present), BIOSIS (1926 to present), and the Cochrane Central Register of Controlled Trials via Ovid (1960 to present). The search strategy incorporated Boolean logic, controlled vocabularies (for example, Medical Subject Heading terms) and free-text words. Because the panel was focused on determining the appropriateness of PICC use in hospitalized adults, articles that included only pediatric patients or devices not comparable with PICCs (for example, arterial or hemodialysis catheters) were excluded.

We also included relevant guidelines, such as the Infusion Nursing Society Standards of Practice (32), Centers for Disease Control and Prevention/Healthcare Infection Control Practices Advisory Committee central line–associated bloodstream infection prevention guidelines (33), American Society of Anesthesiology Task Force on Central Venous Access (34), American College of Chest Physicians Antithrombotic Therapy and Prevention of Thrombosis Guidelines (35), and International Clinical Practice Guidelines for the Treatment and Prophylaxis of Thrombosis Associated With Central Venous Catheters in Patients With Cancer (36).

All retrieved articles were independently scanned for eligibility by 2 of the authors. Disagreements on eligibility were resolved by consensus, and a final list of eligible studies and tables summarizing the evidence were created. The search strategy is provided in Appendix Table 1, and Table 1 summarizes the included articles.

Appendix Table 1. Literature Search Strategy

Table 1. Summary of Studies Included in the Literature Review

Table 1–Continued

Table 1–Continued

Table 1–Continued

Table 1–Continued

Participant and Panelist Selection

Viewpoints related to PICC use are known to vary across specialties; thus, what may be appropriate in one field may not be appropriate in another. To foster discussions about these issues, specialists representing vascular access nursing, hospital-based medicine, internal medicine, infectious disease, critical care, nephrology, hematology/oncology, pharmacy, surgery, and interventional radiology were considered necessary to ensure representativeness of the panel. Leading national and international experts from each of these professions who are eminent scholars or researchers, represent relevant medical societies, or have substantial clinical experience in the field were invited to participate.

To ensure that deliberations took into account patient-centered viewpoints, we also invited a patient to participate on our panel. We recognized that the ideal patient had to be able to speak about experiences with PICCs and related VADs. We recruited such a patient from our university practice in Ann Arbor, Michigan. Owing to the scientific nature of the material, however, the patient panelist did not rate scenarios and instead contributed to panelist discussions. Through this process, 15 multispecialty panelists were recruited to develop the Michigan Appropriateness Guide for Intravenous Catheters (MAGIC) (Appendix Table 2,).

Appendix Table 2. Characteristics of MAGIC Panel Members

Creation of Scenarios

On the basis of articles found through the systematic literature searches, we created clinical scenarios to rate the appropriateness of insertion, maintenance, and care of PICCs. To accurately reflect clinical decision making, devices, including peripheral intravenous catheters, ultrasonography-guided peripheral intravenous catheters, midline catheters, nontunneled CVCs, tunneled CVCs, and ports, were compared with PICCs (Figure 1). Scenarios were crafted so as to allow judgment of real-world use of PICCs; thus, areas of consensus, controversy, and ambiguity were purposefully included. To further ensure validity, we asked each expert to provide a list of concerns related to PICC use that were most relevant to their practice (Appendix Table 3). If not already represented, these issues were also incorporated into scenarios of appropriateness.

• Download figure
• Download PowerPoint

Appendix Table 3. Sample Lists of Thematic Concerns Raised by Panelists*

We developed a conceptual framework to ensure that scientific content, clinical indications, relevant VADs, and contextual factors were adequately represented when drafting scenarios (Figure 2). Thus, indications for PICC insertion were systematically categorized into 1) duration of venous access (≤5 days, 6 to 14 days, 15 to 30 days, ≥31 days); 2) type of infusate (for example, irritants or vesicants, including parenteral nutrition and chemotherapy); and 3) use for specific reasons, such as frequent obtaining of blood samples, poor or difficult venous access, and continuation of intravenous therapies in the outpatient setting. For each of these instances, clinical scenarios incorporating 1) patient-specific factors (for example, critical illness, cancer diagnosis, stage of chronic kidney disease [CKD]), 2) device-specific factors (number of lumens, gauge, type of PICC, alternative VADs), and 3) provider-specific factors (the operator inserting the PICC, technique for PICC insertion) were created. In addition, scenarios regarding appropriate practices for care, management, and treatment of PICC complications were written. Finally, because lack of peripheral access often prompts PICC use for specific clinical needs (for example, need for contrast-based studies or blood transfusion), scenarios related to use of peripheral intravenous catheter in such settings were created.

• Download figure
• Download PowerPoint

We pilot-tested all scenarios with 2 hospital-medicine physicians and further edited them for content and clarity on the basis of their feedback. In this manner, 665 scenarios and 391 unique indications for PICCs and related VADs were developed.

Rating Process

Rating of scenarios and indications were conducted over 2 rounds. In round 1, each panelist received the literature review, definitions of all terms used, a rating document, and instructions for rating. Panelists were asked to dedicate at least 4 hours to complete the rating document. In accordance with the RAND/UCLA method, panelists were instructed not to consider cost when making judgments; rather, they were asked to use the available scientific evidence and best clinical judgment in rating appropriateness (Supplement,). To ensure that appropriateness was rated exclusive of confounding circumstances (such as specialist availability), panelists were also instructed to assume availability of all resources related to the scenarios.

For each indication, panel members rated appropriateness by considering the benefit–harm ratio on a scale of 1 to 9, where 1 indicated that harms outweigh benefit and 9 signified that benefits outweigh harm; Appendix Table 4 provides examples of this process. A middle rating of 5 signified that harms or benefits were equal, or that the rater could not make an informed judgment on the indication. For a series of indications where 2 devices were appropriate, we asked panelists to rate preference for use of one device compared with the other, regardless of cost. Median ratings on opposite ends of the scale (for example, 1 to 3 or 7 to 9) were used to indicate preference of one device over another; a rating in the range of 4 to 6 suggested no preference.

Appendix Table 4. Example Scenarios From Ratings Material

Each panelist rated every scenario twice in a 2-round, modified Delphi process. In the first round, ratings were made individually and no interaction between panelists occurred. In the second round, panel members traveled to Ann Arbor, Michigan, for an in-person meeting where individualized documents showing their ratings along with the distribution of all first-round ratings of the panel were provided.

Over 2 days, a RAND/UCLA methodology expert and a scientific content expert moderated a panel discussion of all indications and scenarios. The sessions were structured to encourage debate and discussion specifically about ratings where disagreement (opposite ratings) or neutrality/uncertainty (ratings of 4 to 6) occurred in round 1. For instance, it often became apparent in the second round that panelists had disagreed not on the indication, but on the patient or circumstances being considered because of inherent assumptions, specialty-specific views, or ambiguity in the scenario itself. When this occurred, the scenario was rewritten with input from the entire panel such that clarifying language or necessary specification was included.

For example, ratings for PICC insertion in patients with CKD were found to be widely disparate in round 1. During round 2, our panel nephrologist clarified that placement of PICCs in patients with stage 3b or greater CKD was specifically contraindicated. Therefore, for indications that included CKD, 2 sets of scenarios were created (stage 3a or lower vs. stage 3b or higher), using Xs and Os on the rating form to distinguish these ratings. Panelists then rerated each of the scenarios, improving validity and agreement of their responses.

Data Processing and Analysis

First-round ratings were submitted either electronically via an online survey system or through paper forms. Data obtained from paper ratings were manually entered into a study database (Qualtrics Research Suite Package, Qualtrics USA) and checked in duplicate for transcription errors. Descriptive statistics (mean, median, mode) were calculated for all variables. A summary result document was created that listed the frequency of responses, median responses, and each individual panelist's response for every scenario. In accordance with the RAND/UCLA method, all indications were classified into 3 levels of appropriateness:

1. Appropriate: panel median score of 7 to 9, without disagreement;

2. Uncertain/neutral: panel median score of 4 to 6, or with disagreement regardless of median; and

3. Inappropriate: panel median score of 1 to 3, without disagreement.

Disagreement was said to have occurred when at least 5 of the 15 panel members rated an indication as appropriate (median score, 7 to 9) and at least 5 panelists rated the same indication as inappropriate (median score, 1 to 3). Only indications without disagreement were classified as inappropriate or appropriate.

Definitions

To ensure consistency, standardized definitions of devices (for example, PICC, midline), populations (active cancer, "special" populations), indications (for example, frequent obtaining of blood samples, hemodynamic monitoring), and infusates (irritant, vesicant) were provided to panelists. A complete glossary of terms and definitions used is provided in the ratings document in the Supplement.

Role of the Funding Source

This project was supported by a Young Researcher Award from the Society of Hospital Medicine to Dr. Chopra. Funds were used to support panelist lodging, meals, transportation, and venue. Blue Cross Blue Shield of Michigan provided salary support for 3 of the authors through a grant to the University of Michigan. Neither funder had a role in the design, conduct, or analysis of the project or the decision to submit the manuscript for publication.

Results

Within the 665 scenarios reviewed, panel members rated 391 unique indications for PICCs and related VADs. During the first round, the panel rated 237 scenarios as appropriate (36%), 267 as inappropriate (40%), and 161 as neutral/uncertain (24%). After the second round of in-person interactions, 253 scenarios were rated as appropriate (38%), 288 as inappropriate (43%), and 124 as neutral/uncertain (19%). Thus, during the second round of discussions, better distinction of neutral/uncertain indications as being appropriate or inappropriate indications occurred. A substantial proportion of this convergence in ratings reflected resolution of disagreement (30 of 37 scenarios) from round 1 to round 2.

1. Appropriateness of PICC Insertion in Specific Populations

A. Appropriateness of PICC Insertion in Hospitalized Medical Patients

In hospitalized medical patients, panelists rated insertion of PICCs for infusion of peripherally compatible infusates as inappropriate if the expected duration of use was 5 or fewer days. In such scenarios, use of peripheral intravenous catheters or ultrasonography-guided peripheral intravenous catheters was rated as appropriate.

If the proposed duration of infusion was 6 to 14 days, panelists rated PICC use as appropriate but indicated a preference for midline catheters and ultrasonography-guided peripheral intravenous catheters over PICCs for this period. This rating reflected evidence from observational studies that suggested both efficacy and lower risk for complications associated with these devices compared with PICCs for this interval (37–41).

When the proposed duration of infusion was 15 or more days, PICCs were preferred to midline catheters, given the possibility of failure of the latter beyond this period (42, 43). However, panelists recognized that midline catheters may be used for up to 4 weeks and are approved for such duration of use (32).

Use of tunneled catheters and implanted ports were rated appropriate only if the proposed duration of infusion was 31 or more days. Panelists noted that these more invasive devices should be reserved for instances when use of PICCs is not feasible (for example, no suitable vein or site of insertion for PICC is identified), is relatively contraindicated (for example, recent history of thrombosis), or when episodic infusions over several months are necessary (Figure 3).

• Download figure
• Download PowerPoint

For infusion of irritants or vesicants, such as parenteral nutrition or chemotherapy, PICC use was rated as appropriate at any proposed duration of use. Because peripheral intravenous catheters, ultrasonography-guided peripheral intravenous catheters, and midline catheters would not provide central venous access, these VADs were rated as inappropriate for this indication for all durations of use.

If skilled operators are available, panelists rated use of nontunneled CVCs as appropriate when the expected duration of use was 14 or fewer days. Panelists also rated use of tunneled, cuffed catheters and implanted ports as appropriate for infusion of irritants or vesicants, but only when the proposed duration of therapy was 15 or more days or 31 or more days, respectively (Figure 4).

• Download figure
• Download PowerPoint

Panelists disagreed on the appropriateness of PICC placement when the indication was frequent obtaining of blood samples (≥3 phlebotomies per day) or difficult or poor peripheral venous access for proposed durations of 5 or fewer days. Our patient panel member actively participated in this discussion, suggesting that such decisions should be individualized between the patient and provider after discussing risks and benefits related to PICC use and alternative options. Insertion of PICCs was rated as appropriate when the proposed duration of use for frequent phlebotomy or difficult venous access was 6 or more days. In patients with difficult venous access, ultrasonography-guided peripheral intravenous catheters and midline catheters were preferred over PICCs when the expected duration of use was 14 or fewer days. Panelists rated use of CVCs for both difficult venous access and frequent phlebotomy as appropriate, provided the proposed duration of use was 14 or fewer days. Placement of tunneled catheters for patients with difficult venous access was rated as appropriate only if the proposed duration of use was 31 or more days. Ports were rated as inappropriate for frequent obtaining of blood samples at all durations and appropriate for difficult venous access if use for 31 or more days was expected (Figures 5 and 6).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

B. Appropriateness of PICCs in Patients With CKD, Cancer, or Critical Illness

Panelists rated the appropriateness of PICC placement in patients with CKD according to disease stage as defined by the Kidney Disease: Improving Global Outcomes CKD Work Group (44). Among patients with stage 1 to 3a CKD (estimated glomerular filtration rate ≥45 mL/min), rating of indications for PICC use followed those of general medical patients. However, the panel noted that managing such patients on the basis of CKD stage alone might be imperfect because myriad factors, including age, magnitude of albuminuria, race, and blood pressure, influence progression of renal disease (45–49). The panel therefore recommended consultation with a nephrologist before PICC insertion if ambiguity regarding the severity of underlying kidney disease exists. However, for patients with stage 3b CKD or greater (estimated glomerular filtration rate <45 mL/min), panelists acknowledged the imperative to preserve peripheral and central veins for possible hemodialysis or creation of arteriovenous fistulae and grafts (49). Thus, regardless of indication, insertion of devices (PICCs, midline catheters) into arm veins was rated as inappropriate in such patients. When venous access for 5 or fewer days was necessary, panelists recommend placement of peripheral IVs in the dorsum of the hand (avoiding the forearm veins) for infusion of peripherally compatible infusates. If venous access for longer durations or infusion of a non–peripherally compatible drug is needed, use of tunneled small-bore central catheters (for example, 4-French single-lumen or 5-French double-lumen catheters inserted in the jugular vein and tunneled toward the chest) was rated as appropriate (50). For patients receiving any form of renal replacement therapy, panelists also recommended consultation with a nephrologist to discuss the possibility of drug administration during or toward the end of the dialysis procedure.

These recommendations notwithstanding, panelists acknowledged that recommendations for patients with stage 3b CKD or greater would need to be individualized, taking into account such factors as the urgency of the situation; rationale for venous preservation; likelihood of eventual renal replacement therapy; and availability of resources, such as tunneled small-bore central catheters.

Given the risks for and consequences of infectious (51, 52) and thrombotic (53–55) complications, as well as the unique indication of chemotherapy, ratings for PICC placement in patients with cancer differed from those for general medical patients. Recognizing the heterogeneity of thrombosis risk in patients with cancer, the panel discussion focused largely on patients with solid tumors. Panelists debated on whether ratings for chemotherapy should be structured by cycles of treatment versus time; given the desire for generalizability, the panel agreed on time as a more practical scale. Therefore, for infusion of nonirritant or nonvesicant chemotherapy, PICCs were rated as appropriate only if the proposed duration of such treatment was 3 or fewer months.

When peripherally administrable chemotherapy for less than 3 months was necessary, panelists disagreed on PICC appropriateness, given the availability of high-quality evidence regarding risk for thrombosis with these devices in patients with cancer (16). However, members of the panel cited conflicting evidence regarding nonthrombotic complications associated with PICC use (15, 56–58). Of note, a study published since the panel meeting (coauthored by one of our panelists) reported a low rate of PICC complications when proper care was ensured (59). Nevertheless, given the divergent data, panelists rated interval placement of peripheral intravenous catheters with each chemotherapy treatment as the most appropriate strategy.

Like PICCs, tunneled, cuffed catheters were rated as appropriate when at least 3 months of treatment were proposed or when PICCs were not feasible (for example, peripheral veins were not available). Ports were rated as appropriate if the duration of treatment was projected to be 6 or more months, but neutral for durations of 3 to 6 months. Panelists noted that earlier use of ports may be appropriate but may be challenging owing to coagulation abnormalities or availability of interventional radiology.

For infusion of irritant or vesicant chemotherapy, panelists rated PICC or tunneled, cuffed catheter use as appropriate at all time intervals; ports were rated as neutral at 3 to 6 months and appropriate at 6 or more months. Panelists recommended tunneled, cuffed catheters over multilumen PICCs in settings where multiple or frequent infusions are required, citing lower risk for complications (60). However, panelists preferred PICCs to tunneled, cuffed catheters when managing patients with coagulopathy and those with severe or prolonged thrombocytopenia (61). When the indication for PICC placement was frequent phlebotomy or difficult peripheral venous access in a hospitalized patient with cancer, panelists raised the threshold for PICC use compared with general medical patients. Thus, PICCs were considered appropriate only if the proposed duration of use was 15 or more days; midline catheters were rated as appropriate for 14 or fewer days of use.

Appropriateness of indications for PICC insertion in critically ill patients also differed from those for general medical patients, given the likely availability of intensivists who could insert CVCs and concerns about hemodynamic stability, infection, and thrombosis. Panelists consequently rated PICC use as inappropriate for infusion of peripherally compatible infusates unless the proposed duration of treatment was 15 or more days. For the same indication, peripheral intravenous catheters and midline catheters were rated as appropriate for proposed durations of 5 or fewer days and 6 to 14 days, respectively. Although limited data supporting the recommendation for midline catheter use in critical care patients were available at the time of the meeting, a recent study reported favorable outcomes and cost savings with this device (62). Central venous catheters were rated as appropriate when the proposed duration of treatment was 6 to 14 days in hemodynamically stable patients; use of CVCs for proposed durations beyond 15 days was rated as uncertain, with panelists expressing concerns about infection and thrombosis.

In hemodynamically unstable patients or scenarios where invasive hemodynamic monitoring or central access was necessary, insertion of CVCs and PICCs was rated as appropriate for durations of 14 or fewer days and 15 or more days, respectively. Panelists preferred CVCs to PICCs in patients who were hemodynamically unstable or were actively receiving vasopressors. In this setting, urgent requests for PICC placement were rated as inappropriate. Given the risk for insertion complications, panelists preferred use of PICCs to CVCs in critically ill patients with coagulopathies (such as disseminated intravascular coagulation or sepsis), especially if use for more than 15 days was proposed.

C. Appropriateness of PICC Insertion in Special Populations

Panelists rated the appropriateness of PICCs in populations that need lifelong intravenous access (for example, sickle cell anemia, short-gut syndrome, or cystic fibrosis) and populations residing in skilled nursing facilities.

For populations that may require lifelong access, ratings were structured on the basis of how often patients may be hospitalized within 1 year. For patients who are infrequently hospitalized (≤5 hospitalizations per year), PICC insertion was rated as inappropriate when the expected duration of use was 5 or fewer days. Insertion of a PICC was rated as uncertain when the expected duration of use was between 6 and 14 days. The panel preferred midline catheters to PICCs for this duration, assuming that peripherally compatible infusates were proposed (63). However, PICCs were rated as appropriate when the duration of use was expected to last 15 or more days.

More permanent devices, such as tunneled, cuffed catheters or ports, were not considered appropriate for patients with infrequent hospitalizations, but our patient panelist (reflecting on her experiences) commented that an individualized approach would be necessary in such situations. In contrast, when patients in this category are frequently hospitalized (≥6 hospitalizations per year), panelists rated use of tunneled, cuffed catheters as appropriate when the expected duration of venous access was 15 or more days. Ports were rated as appropriate when the proposed duration of use in frequently hospitalized patients was expected to be 31 or more days. Panelists preferred placement of tunneled, cuffed catheters to PICCs when use for 15 or more days was expected, citing the need to preserve veins to meet future, likely recurrent needs.

For patients residing in skilled nursing facilities, PICCs were rated as appropriate for infusion of nonirritant, nonvesicant treatments or frequent phlebotomy if the proposed duration of use was expected to be more than 15 days. Appropriateness of PICC was rated as uncertain for durations of 6 to 14 days, where panelists rated midline catheters as appropriate. For venous access of 5 or fewer days, peripheral intravenous catheters were rated as being the most appropriate VAD. Given the variable resources in such facilities and challenges in obtaining venous access, the appropriateness of midline catheters was rated neutral for this period. For infusion of irritants or vesicants in this setting, panelists rated PICCs as appropriate regardless of duration of use.

A summary of these ratings is provided in Table 2.

Table 2. Guide for PICC Use

2. Appropriateness of PICC Practices

A. Appropriateness of PICC Insertion Practices

Before PICC insertion for specialty-specific indications, panelists rated consultations with specialists as appropriate (for example, infectious diseases before placement of a PICC for intravenous antibiotic therapy, or hematology–oncology before PICC insertion for chemotherapy). For patients who require prolonged antibiotic infusions (for example, infections, such as osteomyelitis), panelists rated PICC placement within 2 to 3 days of hospital admission as appropriate in the absence of bacteremia. In the presence of bacteremia, PICC placement was rated as uncertain owing to ambiguities regarding pathogen, intensity of bacteremia, and clearance of infection, among other factors. Consultation with infectious diseases specialists was suggested in this setting.

Preferential placement of PICCs by interventional radiology professionals was rated as appropriate when 1) a suitable target vein for insertion cannot be identified on bedside ultrasonography, 2) the guidewire or catheter fails to advance during bedside placement, or 3) the patient requests sedation that cannot be safely delivered at the bedside. In addition, placement by an interventional radiologist was rated as appropriate for patients with bilateral mastectomy, altered chest anatomy, or superior vena cava filters. For patients with permanent pacemakers or defibrillators, preferential placement by an interventional radiologist rather than a vascular nursing professional was rated as appropriate if the contralateral arm was not amenable to insertion. These ratings were largely driven by expert opinion.

Panelists rated the appropriateness of specific PICC insertion practices on the basis of availability of the contralateral arm for placement. In accordance with Infusion Nursing Society Standards of Practice (32), avoiding insertion over a bruised or corded venous segment, near or over an open wound or burn, and into veins below the elbow was rated as appropriate. Owing to heightened risk for thrombosis, panelists rated avoiding PICC placement in a hemiparetic or immobile arm as appropriate when the opposite limb was available (64). Avoiding PICC insertion in the dominant arm as a strategy to prevent complications was rated as inappropriate, given the lack of convincing data to support this practice. However, our vascular nursing and patient panelists recommended that technical aspects and patient preferences be considered when selecting arm of insertion.

Prior to PICC use, radiographic verification of PICC tip position was rated as appropriate after blind bedside PICC placement or admission to a hospital with an existing PICC. Conversely, panelists rated routine radiographic verification of PICC tip position as inappropriate when PICCs were placed with electrocardiographic guidance, provided that proficiency with this technology had been demonstrated and adequate tracings (such as P-wave deflections) were observed.

To limit the risk for thrombosis, the U.S. Food and Drug Administration and specialty societies recommend that CVCs terminate in the lower one third of the superior vena cava or cavoatrial junction; "higher" (such as the upper one third of the superior vena cava) or "lower" positions (such as the right atrium) were not recommended (32, 65, 66). Acknowledging these concerns, panelists rated adjustment of the PICC when the tip was in the upper or middle one third of the superior vena cava or right ventricle as appropriate.

However, panelists deviated from existing recommendations in rating the right atrium as an appropriate position for the PICC tip and one that does not warrant adjustment. This rating was made after extensive discussions of clinical practice and review of contemporary evidence, which did not suggest that termination of PICCs or CVCs in the right atrium was associated with adverse outcomes in adults (66–71). Panelists recognized that supporting data were observational, and a well-conducted randomized, controlled trial would be helpful in supporting this recommendation.

The possibility of atrial tachyarrhythmia during or after PICC insertion in this position was also debated (72). As with any CVC, placement of the PICC tip in the right atrium in the setting of an atrial arrhythmia was not recommended. However, in the absence of contraindications, repositioning the PICC tip simply because it resides in the right atrium was rated as inappropriate.

B. Appropriateness of PICC Selection, Care, and Maintenance Practices

Without a documented rationale for a multilumen PICC (for example, multiple incompatible fluids), panelists rated default use of single-lumen devices as an appropriate and potentially important way to reduce PICC complications (73–75). Insertion of multilumen PICCs to separate obtaining blood samples from giving infusions or to ensure a "backup" lumen was available was also rated as inappropriate. To clarify device needs, collaboration with pharmacists or vascular access operators before ordering a PICC was rated as appropriate.

Regarding dressings, panelists rated placement of sterile gauze between the PICC entry site and adhesive dressing for the first 1 to 2 days of insertion as appropriate; thereafter use of clear, transparent dressings that permit site examination and weekly or more frequent changes of wet, loose, or soiled dressings was rated appropriate. Use of cyanoacrylate products ("super glue") to prevent oozing or discharge from the exit site or to secure catheters was rated as neutral by panelists, who noted lack of substantial evidence or experience to support this recommendation (76). In accordance with available guidelines (33), routine use of chlorhexidine dressings without documented adherence to basic infection-prevention efforts or in the absence of high rates of central line–associated bloodstream infection was rated as inappropriate.

Panelists rated use of normal saline rather than heparin to maintain catheter patency and prevent lumen occlusion as appropriate, as reflected in recent recommendations (77, 78). Regardless of how far out the PICC was dislodged, panelists rated advancement of migrated PICCs as inappropriate; in this setting, guidewire exchange of the PICC was rated as appropriate, provided that there are no signs of local or systemic infection. Guidewire exchange was also rated as appropriate when changes to existing PICC characteristics (such as number of lumen or power-injection compatibility) were desired. Should a PICC no longer be functional, exchange over a guidewire was rated as appropriate, provided that an indication warranting continued PICC use was present. Ratings regarding guidewire exchanges were driven largely by expert recommendation.

C. Appropriateness of Management of PICC Complications

In patients with a centrally positioned, otherwise functional PICC that is complicated by image-confirmed PICC-related deep venous thrombosis (DVT), panelists rated PICC removal as appropriate only when 1) the PICC is clinically no longer necessary; 2) the PICC is only being used for phlebotomy, but peripheral veins are available; 3) symptoms of venous occlusion (arm pain, swelling) persist despite therapeutic anticoagulation for 72 or more hours; and 4) bacteremia with objective evidence of line-related infection exists. Panelists rated removal of a functional PICC in the presence of DVT as inappropriate when 1) irritants or vesicant infusions remain necessary; 2) the patient has poor peripheral venous access and requires frequent phlebotomy (and may thus require another PICC); and 3) the patient has minimal improvement in symptoms of venous occlusion, but therapeutic anticoagulation has been provided for 72 or fewer hours. Panelists were neutral regarding PICC removal when 1) a patient could not receive systemic anticoagulation, but the PICC remained clinically necessary and 2) a line-related infection was suspected, but not confirmed. In general, these ratings mirrored existing evidence-based recommendations (35, 53, 79).

When treating PICC-related DVT, panelists rated provision of at least 3 months of anticoagulation at a treatment dose as appropriate. Shorter durations of anticoagulation or removal of the PICC as definitive therapy (in the absence of contraindications to anticoagulation) was rated as inappropriate. When treating with warfarin, panelists recommended targeting anticoagulation to an international normalized ratio of 2 to 3; lower or higher international normalized ratio targets were rated as inappropriate. Use of low-molecular-weight heparin over warfarin was preferred in patients with cancer. Owing to insufficient evidence, preferential use of target-specific oral anticoagulants over traditional agents among patients with cancer was rated as inappropriate. Panelists rated urgent referral to interventional radiology for catheter-directed treatment of PICC-related DVT as appropriate when symptoms of venous occlusion were associated with phlegmasia cerulea dolens (swollen, enlarged, painful, and purplish discoloration of the affected limb).

Panelists rated the appropriateness of placement of a new PICC in patients who experienced PICC-related DVT within the past 30 days. In this scenario, panelists strongly urged against placement of a PICC, given the high risk for recurrent thrombosis. Placement of a PICC was specifically rated as inappropriate if the indication for insertion was 1) frequent phlebotomy when peripheral access was available, or 2) patient request for comfort in non–end-of-life settings. Insertion of a PICC was also considered inappropriate if the patient were to require surgery lasting 1 hour or longer, owing to heightened risk for DVT in this situation (67).

In the setting of PICC-related DVT, appropriateness of PICC insertion for parenteral antibiotics for 10 or more days was rated as uncertain; panelists recommended a midline catheter in this scenario. If a PICC was absolutely necessary in a patient with recent PICC-related DVT, panelists rated use of the smallest catheter gauge and least number of lumens as appropriate (74, 75, 80). Placement in a vein in the contralateral arm following at least 3 months of anticoagulation for the PICC-related DVT was also rated as appropriate in this setting.

Panelists rated the appropriateness of practices related to management of PICC-related bloodstream infections. Regardless of clinical context and in accordance with recommendations (33, 81), panelists rated use of PICCs as a strategy to reduce bloodstream infection as inappropriate. In the setting of bacteremia or fever, PICC removal in the absence of confirmatory evidence of line-related infection was rated as uncertain. Panelists stated that these approaches would be dictated by such factors as pathogen, intensity of bacteremia, and clinical stability, among others, and consultation with infectious disease would be appropriate.

In patients with confirmed PICC-related bloodstream infection, continued treatment using the affected PICC, guidewire exchange, or placement of a new device in the contralateral arm without documented clearance of infection was rated as inappropriate. After a line-free interval (typically 48 to 72 hours) and negative blood cultures, panelists rated placement of a PICC or other acute CVC as appropriate only if an indication warranting central catheter use was present. Panelists preferred use of peripheral IVs in such patients wherever possible.

D. Appropriateness of PICC Removal

In contradistinction to indwelling urinary catheters (82), panelists rated PICC removal without physician notification as inappropriate. After physician notification, panelists rated PICC removal as appropriate when 1) the PICC has not been used for any clinical purpose for 48 hours or longer; 2) the patient no longer has a clinical indication for a PICC, or the original indication for use has been met (for example, an antibiotic course has been completed); or 3) the PICC is only used for routine obtaining of blood samples in a hemodynamically stable patient and peripheral veins are available. Panelists rated routine removal of a PICC in a hemodynamically stable patient with poor venous access or hemodynamically unstable patients as uncertain. Removal of a PICC by clinicians who have received training to remove CVCs, but not PICCs, was rated as inappropriate (32).

A summary of these ratings is provided in Table 3.

Table 3. Guide for PICC Insertion, Care, and Maintenance Practices

3. Appropriateness of Peripheral Intravenous Catheter Use in Specific Scenarios

Because PICC use is often driven by difficult peripheral venous access, we asked panelists to rate appropriateness of peripheral intravenous catheter use in various clinical scenarios that often prompt PICC use. In the absence of other indications for central venous access, panelists rated use of ultrasonography-guided peripheral intravenous catheters as appropriate before insertion of a PICC in general medical, critically ill, and cancer populations with difficult venous access (39, 68). However, use of ultrasonography-guided peripheral intravenous catheters in patients with stage 3b or greater CKD was rated as inappropriate. If a suitable arm vein could not be found, panelists rated placement of a peripheral intravenous catheter catheter in the external jugular vein of the neck as appropriate only if the proposed duration of use was 96 hours or less or in an emergency situation. Panelists rated placement of a peripheral intravenous catheter in the lower extremity as appropriate only in emergencies.

Citing the results of a Cochrane systematic review (83) and a randomized trial (84), panelists rated replacement of peripheral intravenous catheters as appropriate when prompted by clinical signs and symptoms rather than prespecified durations. Panelists noted that such practice might extend availability of peripheral venous access (83), reduce cost (85), and limit use of PICCs, but recognized that these data were limited to 1 randomized trial and low event rates in the literature. When PICC placement was requested for blood transfusions, panelists rated 16-, 18-, and 20-gauge peripheral intravenous catheters as appropriate and preferable to PICC use. For administering intravenous contrast through radiographic injectors, panelists rated use of 16- to 20-gauge peripheral intravenous catheters as appropriate and preferred over PICCs; use of 22-gauge devices or larger was rated as inappropriate.

A summary of these ratings is provided in Table 4.

Table 4. Guide for Peripheral Intravenous Catheter Practices

Discussion

Our 15-member multidisciplinary panel successfully applied the RAND/UCLA Appropriateness Method to generate novel criteria for use, care, and management of PICCs in hospitalized patients. In addition, panelists rated the comparative utility of other VADs in relation to PICCs, providing new insights for decision making in venous access. The implication of this work is substantial, because it provides a potential means to quantify appropriateness, qualify existing use, and improve care of PICCs and related devices in hospitalized patients. Given an international team of experts that represented multiple subspecialties and the inclusion of a patient to formulate panelist ratings, these criteria are well-positioned to broadly improve the quality and safety of venous access in hospitalized adults.

As with many health care innovations, PICCs were introduced to solve an important clinical problem in a defined population (86). However, over time, the use of PICCs has evolved to span diverse indications and patient populations. In hospital settings, accumulating evidence suggests that placement of PICCs may occur for potentially inappropriate reasons (18, 87). Notwithstanding such benefits as convenience, comfort, and economic efficiency (4, 88), PICC insertion may introduce unnecessary risk and potential for preventable harm (15, 16, 73). Despite this fact, no framework to inform use of these devices has been developed to date.

These observations were the motivation underlying this project, which sought to incorporate existing evidence with the knowledge of clinicians and content experts to define criteria for appropriate PICC use. Unlike existing recommendations, our appropriateness criteria represent a departure from the status quo in several ways.

First, they offer clinical granularity for clinicians. For example, existing guidelines recommend "use of midline catheters or PICCs instead of a short peripheral intravenous catheter when the duration of IV [intervenous] therapy will likely exceed six days" (33). Our criteria build on this advice by adding such details as what patient-specific considerations should be incorporated in this decision, which other devices may be appropriate, and when PICC use for shorter durations might be reasonable.

Second, whereas existing recommendations target proceduralists or specialties that most often insert devices, our criteria are the first to provide direction to clinicians, such as internists or hospitalists, who order PICCs. Thus, these criteria fill a critical gap, bringing recommendations to those that drive the decision to place such devices.

Finally, by tackling some of the most controversial topics of venous access—including when to adjust the PICC position, appropriate indications for removal, and indications for reinsertion of PICCs after complications—our criteria advance the science of vascular access in important and innovative ways.

Some aspects of panelist deliberations and ratings merit further discussion. First, patterns of recommendations for PICC appropriateness often hinged on 2 variables: the nature of the infusate and duration of venous access. Thus, non–peripherally compatible infusions or scenarios where venous access was necessary for 6 days or longer often led panelists to rate PICC use as appropriate; conversely, shorter duration of use with peripherally compatible infusions led to a recommendation for use of a peripheral intravenous catheter, ultrasonography-guided catheter, or midline catheter. Unlike existing standards, however, variation in risk for complications according to patient population influenced this pattern. This is well-illustrated in ratings for critically ill patients and those with cancer, where a theme of limiting PICCs to durations of use of 15 days or longer is evident.

Second, throughout deliberations, panelists noted that it is often challenging for clinicians to estimate an expected duration of venous access. Relatedly, a "maximal" window within which PICCs may be safely used is not known and depends on myriad factors, including adequacy of care and differential risk for complications. Finally, panelists acknowledged that separation of indications for PICC placement into individual categories and defining VADs by finite duration was artificial, because venous access is rarely driven by a single clinical purpose or limited by duration.

On balance, panelists rationalized that clinicians often do not reflect carefully enough on the nature of venous access or weigh its inherent risks and benefits. Panel members added that in many hospitals, the decision to place a PICC is often dichotomous, with consideration of other devices lacking. Thus, an unforeseen advantage of these criteria is the introduction of a physician-directed "time-out" in vascular access decision making. During this pause, reflection on the appropriate device, patient risk factors, and discussions with specialists could conceivably improve outcomes in hospital settings.

Our approach has several limitations. First, we excluded neonatal and pediatric studies when formulating these recommendations, because considerable differences in PICC use exist between these patients and adults. However, because these populations often receive PICCs, future panels should choose to focus on these subsets.

Second, although our panel was multidisciplinary, we did not include bedside nurses, who often request PICCs in hospitalized settings. However, vascular nurses and hospitalists are attuned to considerations regarding PICC use from this group of providers and were well-represented on our panel.

Third, the applicability of these recommendations will vary on the basis of provider scope of practice, education, and training. As echoed in other standards (89), provider availability, competence, and technical expertise should guide insertion and selection of appropriate VADs.

Finally, our panel was focused on appropriateness of PICCs in relation to other devices. We acknowledge that certain devices may be used for longer durations (for example, midline catheters for up to 28 days) or indications of different durations (for example, intravenous antibiotics for 6 weeks). These limitations were necessary to ensure comparability among various devices and generalizability of these recommendations.

Despite these limitations, our appropriateness criteria represent a major multidisciplinary effort toward improving decision making related to PICCs and related VADs. Avoiding PICC use for inappropriate indications, considering alternative devices, ensuring appropriate consultations, and outlining instances where PICC removal is appropriate are but a few examples of how these recommendations may be implemented to improve practice. In addition, by including a patient whose opinion influenced panel deliberations, we took into account the implications of provider decisions from "the other side of the needle." Finally, the criteria we propose span not just indications for PICC insertion but also best practices for use, care, and maintenance. Thus, we hope that our recommendations will provide clarity for management of complex situations not only before, but also during and after, PICC placement.

Although optimal strategies to implement our criteria remain to be defined, an expansive range of options is possible. For example, routine benchmarking and feedback of metrics, such as PICC dwell time, indications for insertion, and practices related to management of complications, may serve to inform hospital-specific "PICC dashboards" and quality-improvement efforts. Alternatively, more sophisticated paradigms, such as decision aids and computerized physician order-entry taking into account proposed duration of use, indication, and patient characteristics, are also plausible.

Because many of our recommendations are algorithmic, Web sites or smartphone applications to determine the appropriateness of PICCs before insertion seem to be feasible. We are beginning to explore these options through 2 strategic partners. First, through the ongoing Blue Cross Blue Shield/Blue Care Network–funded Hospital Medicine Safety collaborative quality improvement project, we will use our appropriateness criteria to evaluate and improve PICC utilization in 47 Michigan hospitals (90). Because the Hospital Medicine Safety project is composed of diverse hospitals and is built on a robust data platform, we will also seek to understand contextual barriers, facilitators, and unintended consequences related to use of our criteria.

Second, through work recently funded by the Veterans Affairs National Center for Patient Safety and the No Preventable Harms Campaign, we will test ways in which to operationalize our criteria within the highly integrated Veterans Affairs health system. Given the advanced electronic medical record systems in this setting, our experiences will shed new light on implementation strategies that could inform our work within and beyond this setting. Such research may take several forms. For instance, quasi-experimental designs, such as pre–post or interrupted time series that examine the influence of specific appropriateness recommendations (for example, avoid use of PICCs for peripherally compatible infusions lasting 5 days or less) within and between hospitals, could be tested in participating Michigan and Veterans Affairs sites. Alternatively, a "bundle" of best practices related to PICCs, including appropriateness criteria for insertion, care, and management, may be deployed, leveraging a step-wedge or cluster randomized approach to account for secular trends.

More robust research designs, such as randomized clinical trials, that utilize our criteria are also feasible. For example, randomly assigning patients who require less than 2 weeks of peripherally compatible infusions to receive a midline catheter or PICC is not only feasible but also relevant, because many PICCs are placed to deliver antibiotics for such intervals after hospital discharge. Such a study may be powered to ascertain the noninferiority of midline catheters, rates of therapy completion, or complications with either device. Therefore, several research designs that span one or more hospitals, and one or more of our recommendations, may be used as interventions to target clinical outcomes, overall utilization, adverse events, and costs.

In conclusion, we used the RAND/UCLA Appropriateness Method to define best practices for PICC insertion, care, and management. Although a key first step, these criteria offer but a blueprint of best practices. To make MAGIC truly happen, diffusion, uptake, and refinement from the providers and stakeholders engaged in vascular access is necessary. Through use of a systematic rating process, a multidisciplinary international panel, and patient representation, we hope to achieve this goal. Our patients deserve nothing less.