Background Peripherally inserted central catheters (PICCs) provide reliable intravenous access for delivery of parenteral therapy. Yet, little is known about PICC care practices or how they vary across hospitals. We compared PICC-related processes across hospitals with different insertion delivery models.
Methods We used a descriptive qualitative methodology and a naturalist philosophy, with site visits to conduct semistructured interviews completed between August 2018 and January 2019. Study sites included five Veterans Affairs Medical Centres, two with vascular access teams (VATs), two with PICC insertion primarily by interventional radiology (IR) and one without on-site PICC insertion capability. Interview participants were healthcare personnel (n=56), including physicians, bedside and vascular access nurses, and IR clinicians. Data collection focused on four PICC domains: use and decision-making process, insertion, in-hospital management and patient discharge education. We used rapid analysis and a summary matrix to compare practices across sites within each domain.
Results Our findings highlight the benefits of dedicated VATs across all PICC-related process domains, including implementation of criteria to guide PICC placement decisions, timely PICC insertion, more robust management practices and well-defined patient discharge education. We also found areas with potential for improvement, such as clinician awareness of PICC appropriateness criteria and alternative devices, deployment of VATs and patient discharge education.
Conclusion Vascular access nurses play critical roles in all aspects of PICC-related care. There is variation in PICC decision-making, care and maintenance, and patient education across hospitals. Quality and safety improvement opportunities to reduce this variation are highlighted.
- nosocomial infections
- patient safety
- qualitative research
Data availability statement
Data are available upon reasonable request. Deidentified, anonymised data that support the findings of this study will be made available by the corresponding author, SLK, on reasonable request and with a letter of agreement.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Reliable intravenous access is critical for delivering vital fluids and medications. Vascular access devices1 are a ubiquitous part of hospital care. Central venous catheters (CVCs) that terminate in large vessels near the heart were once primarily found only in intensive care units (ICUs).2 3 However, peripherally inserted central catheters (PICCs), which are safer to insert than traditional CVCs and enable outpatient parenteral therapy,2 4 are now commonly found inside and outside the hospital setting.
Despite their popularity, information about PICC use, care and management, and patient education within and across hospitals is limited. Yet, these are critical domains for preventing serious complications, such as central line-associated bloodstream infection (CLABSI) and venous thrombosis. Emerging data suggest that implementing appropriateness criteria, which include the use of alternatives such as midlines (devices that do not enter the central veins), can reduce PICC use and PICC-related complications.5 6 Yet, how appropriateness criteria are used by different hospitals is not well described. Studies also suggest that dedicated vascular access teams (VATs) are associated with increased use of CLABSI prevention practices7 and low rates of major complications.8 9 Nonetheless, whether PICC-related care processes differ between hospitals with and without such teams has not been assessed. Finally, despite complication risks,10 11 hospital variation in the education of patients about postdischarge PICC care has yet to be explored.
Data show that VATs insert PICCs in approximately 60% of US hospitals.7 Among hospitals without such teams, over 50% indicate PICCs are inserted by interventional radiology (IR). The number of hospitals without on-site PICC insertion is unknown but likely includes many smaller hospitals with low-volume PICC use. Thus, the objective of this qualitative study was to describe and compare PICC-related processes across hospitals with different PICC insertion models. We were interested in how decisions about PICC use were made, the PICC insertion process, and practices related to in-hospital PICC management and patient discharge education as these are all key domains for ensuring the safe delivery of PICC-related patient care.
Study design and sample
We used a descriptive qualitative methodology and a naturalist philosophy12 which involve studying something as it is in its natural state rather than using a specific theoretical or disciplinary perspective to study a phenomenon. This approach allowed us to develop an in-depth understanding of PICC-related processes as described by key informants in their everyday practice settings at five Department of Veterans Affairs (VA) Medical Centres.
Site visits, primarily to collect data through in-person semistructured interviews, were conducted between August 2018 and January 2019. Sites were purposefully selected based on the PICC insertion model (ie, primarily inserted by VAT, IR or no on-site insertion) to reflect models in use by US hospitals. We identified a point of contact at each site, through colleagues in hospital medicine and infection prevention, who provided lists of potential interviewees involved with PICC insertion and care, including hospital medicine, surgery, critical care and infectious disease (ID) physicians; medical/surgical, critical care and infusion centre nurses; VAT members; and IR clinicians. We invited individuals from these lists, through email and by telephone, to participate in study interviews.
The multidisciplinary research team developed a semistructured interview guide (online supplement file 1) consisting of questions related to our domains of interest: PICC use and decision-making process, PICC insertion, in-hospital PICC management, and patient education and postdischarge care. We focused on these domains, given their link with PICC-related complications and patient safety.6 13–16 Most interviews were conducted in person, by clinician and non-clinician interviewer pairs, during facility site visits of 1–2 days by four members of the research team (SLK, MH, MQ, KEF and VC). Some interviews were conducted by telephone due to scheduling conflicts or because new key informants were identified during the site visit. Two of the interviews were conducted with the hospital point of contact, but otherwise, the interviewees did not know the members of the interview team. Interviews were audio recorded and transcribed verbatim.
Cross-case analysis was conducted with each site considered an individual case study, allowing us to identify similarities and differences across hospitals with different insertion models and organisational features, as well as to identify areas for improvement both within and across hospitals. A rapid analysis approach was selected, given our objectives were to describe and compare PICC-related processes across hospitals, and to identify specific areas, within key domains, for subsequent improvement efforts or further study rather than provide an in-depth examination of more complex phenomenon surrounding PICC use.17 18 We developed a summary template using each process domain represented in the interview guide. The site visit team used the template to create preliminary site summaries after each visit. Next, two other study team members (LEW and BRG) individually reviewed interview transcripts to create detailed summaries for each participant. From the participant summaries, a detailed synthesis incorporating perspectives across professional roles was created for each VA Medical Centre. The interview and preliminary and detailed site summaries were all reviewed and discussed during a series of meetings by the multidisciplinary research team to create a final summary for each site. The site-specific information was then transferred to a matrix to facilitate cross-case comparisons, highlighting similarities and differences across sites, within the four process domains. We established rigour throughout the study by involving researchers from different backgrounds (medicine, nursing and qualitative methods) in data collection and analysis, including discussing our own perceptions and potential biases based on clinical or personal experiences; by comparing similarities within and across interviewees involved in various aspects of PICC use and by looking for discordant evidence.
The characteristics of the five hospitals are shown in table 1. Following our site selection criterion, PICCs were primarily inserted by vascular access nurses at the two largest sites (sites 4 and 5), while PICCs were primarily inserted by IR at sites 1 and 2. However, we learnt while scheduling the site 1 visit that IR was responsible for non-ICU patient PICC insertions, but ICU nurses with specialised training inserted PICCs for ICU patients. At the smallest hospital (site 3, average daily census of 46), PICCs were not inserted on-site. Of the 115 healthcare personnel contacted, 56 consented and were interviewed. A detailed description of interviewees and recruitment is provided in online supplement file 2. Interviews ranged in number from 7 to 14 across sites and from 13 to 59 min in length.
Our main findings within each process domain, including decisions about PICC use, insertion process, in-hospital management and patient discharge education, are discussed further in detail with a focus on notable differences and common challenges across sites.
PICC insertion decision
The PICC insertion decision includes the personnel involved, criteria for determining appropriateness and the availability of alternative vascular access devices. Three primary differences were found across sites as summarised in table 2.
The first difference is the extent to which medical subspecialists were involved in the decision process. At three hospitals, including both with VATs (sites 4 and 5), subspecialists played a role in the insertion decision when specific populations or indications arose, for example, infectious diseases (IDs) when the indication for use was intravenous antibiotics, or nephrology for patients with renal dysfunction. As one vascular access nurse at site 5 described:
We used to stick, put PICC lines in the renal patients, what we're finding is when the patients are finally getting to dialysis, the dialysis catheters, the fistulas are not lasting, they’re crashing faster than ever and I give renal credit because they (renal) really stepped up to put their foot down and make sure those arms are protected.
At sites 1 and 3, subspecialist involvement in the decision process was more variable. At site 1, when asked about involvement in the decision-making process, one ID physician responded:
Yeah, it varies. … early on when I first started the majority of time like PICCs were already going, like in. But now that we’ve had this kind of stronger presence I would say that I think that they’re a little bit better because sometimes they’ll call and then ‘we’re like we’re putting in the order for the PICC line’.
The second difference involved awareness and use of criteria for decisions regarding PICC appropriateness. While awareness of appropriateness criteria varied across sites, only hospitals with VATs reported explicitly using such criteria to guide decision-making. Conversely, at hospitals without a VAT (sites 1, 2, 3), formal criteria to guide decisions did not exist. As described by IR clinicians at site 1:
She’s (ambulatory infusion nurse) really the triage person that says, okay what kind of line … I asked her if she had like a algorithm she follows because we’re doing this at the main (academic) hospital … and she doesn’t. She’s just done this for so long she just sort of knows the flow of things. Consult goes to (ambulatory infusion nurse), she gathers that information. She puts it on the order and then sends it to us. We review it to make sure it’s appropriate. … [Interviewer: And do you use specific criteria to decide when it’s appropriate?] Yeah, well we try to follow the MAGIC guidelines.
Vascular access nurses were using the Michigan Appropriateness Guide for Intravenous Catheters (MAGIC) to determine device appropriateness at both vascular team hospitals. However, ordering clinicians across all sites were often not aware of these criteria except for a few clinicians at site 4, where MAGIC had been incorporated into the electronic order set.
… our order set has been changed where now you have to, and it’s in many of the same categories that are on the MAGIC criteria where you have to click, you know what it is that you’re using that access for. [hospitalist, site 4]
The third difference was the availability and use of potential PICC alternatives (eg, ultrasound guided peripheral intravenous catheters, midlines). Three sites had alternatives available routinely, two did not. Consequently, knowledge about, and perceived utility of alternatives varied across sites. At sites 2 and 3, alternatives were not generally available and ordering clinicians were not familiar with devices such as midlines. However, even at sites with alternatives, placement varied based on perceived usefulness. For example, at site 1, midline availability was not common knowledge and the nurse who reviewed PICC consult requests did not have a favourable opinion about their usefulness. At site five the VAT used midlines occasionally but expressed concerns about functionality. As one site five vascular access nurse described:
… if I go down to bedside and the nurse says, “Let’s put an IV in,” and the doctor will say, “Well, let’s look at the patient for a midline,” I look at them and think, “Okay, if he’s going to be here for the next week and that midline can fail in 2 days because it’s an IV, it’s a long IV,” then I, a lot of times end up jumping over in going to a PICC line.
In contrast, the vascular access nurses at site four were enthusiastic about using midlines or ultrasound guided peripheral intravenous catheters when appropriate.
… the number of PICCs trending down, that’s what I want it to be and the midline is, the midline also is just, just evened out but the peripheral IV, peripheral IV ultrasound guided like goes up like probably triple the number. (Vascular team nurse, site 4)
PICC insertion process
While the type of providers inserting PICCs differed across sites in accordance with our sampling frame, concerns about provider availability and care delays were described at all sites (table 3). These challenges were especially prominent at certain hospitals. For example, at site 2, where only IR clinicians placed devices, PICCs were not inserted after 16:00 or on weekends, and patients with PICCs ordered on a Thursday or Friday could have an insertion delay of 2–3 days.
…we just have interventional radiology that does it. And if they would need it on the weekends good luck because we don’t always have call, we do have call but we don’t, I don’t get it. (Bedside nurse, site 2)
The impact on patient care was even more pronounced at the hospital without on-site insertion (site 3), requiring days for appointment scheduling and an ambulance ride of 30–40 min round trip if transported to a community hospital or 4 hours if transported to the nearest VA Medical Centre.
Finally, while site 4 was expanding on-site VAT availability, site 5 was reducing the team’s on-site presence, with the impact on care delivery noted for both scenarios. According to a hospitalist at site 4:
… they have really expanded the availability of the vascular access team and so I don't know if they’re here 7 days a week, but they’re definitely here one of the days on the weekend, they’re here on holidays, they’re here for an extended period of time and so it has made things so much different because I feel there’s a lot of patients who before there would be, you know the back and forth between nurses and physicians and access is poor and now a lot of them, we’re just calling the vascular access nurses …
In contrast, at site 5:
… in the last month or 2 months, we don’t have anybody on the weekend, we have to call them in. Usually they were on all the time, we had them over the weekend as well and we could put stuff in. (Hospitalist, site 5)
In-hospital PICC care and removal
At all hospitals, registered nurses (RNs) were responsible for in-hospital PICC care. However, which nurses performed which tasks and whether nurses could perform certain tasks differed across hospitals (table 4). At site 1, a dedicated intravenous team assumed sole responsibility for managing PICCs 24/7. At the hospitals with VATs (sites 4 and 5), responsibility for PICC care was shared, with bedside RNs conducting assessments, blood draws and as-needed dressing changes, while vascular access nurses conducted daily assessments, flushes and weekly dressing changes. Vascular access nurses were also responsible for troubleshooting when PICCs were not functioning properly and, if necessary, administering tissue plasminogen activator (tPA) to clear an occlusion and restore patency.
Well, they (bedside RNs) do PRN (as needed), we (vascular access RNs) do dressing changes for the entire hospital every Thursday … But we audit the lines every day. Monday through Friday we go and look at the dressings, we flush the lumens, make sure that they’re patent. If they are not, you know we recommend that they get the Alteplase, have the doctor put the order in and then we’ll de-clot them … (Vascular team nurses, site 4)
At sites 2 and 3, bedside nurses had to contact a physician about any complications and, at site two could administer tPA only with a physician order. At the hospital without on-site insertion capability (site 3), tPA was not available so clotted lines had to be removed.
The process for in-hospital PICC removal also differed across hospitals. While physicians were the primary decision makers regarding removal at all hospitals, at site four the VAT conducted regular reviews and prompted device removal when needed. As described by the vascular access nurses:
every week we have a huddle with our boss (team manager), and then we go over and say like, “Well, there’s a PICC line, do we have medication going on? We don’t have medication going on, if there’s nothing there, it doesn’t qualify to keep it right there,” and sometimes the primary nurses would say that you know patient is a hard stick and we need to draw blood … So, we had to tell them that, “Hey, you don’t go do your home grocery in 18-wheeler, you know?” (all laugh) That’s something like if you’re trying to keep a central line in just to draw blood, it’s pretty much like that, you know.
At sites 4 and 5 vascular access nurses also had primary responsibility for physically removing PICCs, although both hospitals were training bedside nurses to remove PICCs if a vascular access nurse was not available.
then to DC (discontinue) a PICC, we have nurses on the floors that are checked off to do it. We used to do them all because in the past, well-meaning nurses have pulled PICC lines out when patients were going home not realising they were going home with a PICC line and we’ve had to replace them. So, the fix was IV (vascular) team only …, but nowadays if we’re busy the nurses that are checked off can … (Vascular team nurse, site 5)
At site 1, ICU nurses who inserted PICCs were responsible for removal in ICU patients, while dedicated intravenous team RNs had primary responsibility for removal involving non-ICU patients.
At the two sites without a vascular access or intravenous team, physicians played more of a role in the removal process. ICU nurses could remove PICCs for ICU patients at site 2, but only physicians could remove for non-ICU patients. The challenge at site 3, where physicians were responsible for PICC removal, was the lack of physician competency given this was a low frequency event.
nurses are not allowed to remove PICC lines here. It has to be a physician. And it’s frustrating because like if (Hospitalist physician) is on vacation the doctors that come over here don’t know how to do it. … two PICC line people ago I had to walk the doctor (through) how to pull out a PICC line. And like, I can’t do it but you can and you have no clue how to do it. (Bedside nurse, site 3)
Patient discharge education and postdischarge management
We found substantial differences in patient discharge education and post-discharge management as described in table 5. In general, hospitals with VATs (sites 4 and 5) had better-developed discharge teaching approaches. However, there was no standard education or home discharge process across sites. At sites without such teams, bedside, discharge and/or infusion centre nurses were responsible for education. The type of education ranged from an informational handout only (site 3) to educational materials plus skills-based instruction, use of visual aids, and hands-on teaching (sites 1, 4 and 5).
We have PICC education videos that are available on every TV. They’re a few minutes each and one is from, you know what is a PICC line to what, how to take care of a PICC line at home, signs and symptoms, what to do after it’s taken out, you can run them in a series and it’s 15 min total. … We’ll teach them about the flushing, watching for the blood returns, any problems with the PICCs, if the PICC comes out, when to call us, when to come in … (Vascular team nurse, site 5)
Finally, all sites indicated that home care agencies provided care for some, but not all, patients after discharge. Only site four described a process in which the home care nurse came to the hospital prior to patient discharge. According to a nurse case manager at site 4:
…once they received the referral, that (infusion company) nurse will come on-site and she’s going to teach the patient and/or caregiver the initial process … Which is great because that’s a clean handoff…
The home care agency, when provided, was the primary point of contact for patients should they experience PICC problems after discharge. However, at sites 4 and 5, patients were also given contact information for the VAT, whereas intravenous team information was provided at site 1. At site 3, patients were instructed to go to urgent care for device-related issues.
We sought to describe and compare PICC-related processes across hospitals with different PICC insertion models, given PICCs are a vascular access mainstay but require appropriate use and management to reduce the risk of serious complications. Our study characterises PICC-related processes across five diverse hospitals and highlights the critical role of vascular access nurses in the implementation of appropriateness criteria, reducing insertion delays, ensuring robust in-hospital PICC care, and providing discharge education. However, our findings also reveal variation in practice and opportunities within each process domain to improve care delivery and patient safety.
The importance of dedicated VATs for ensuring safe and effective PICC use is not novel. Prior studies show that PICC placement by VATs is associated with fewer insertion complications and can be cost-saving.19–23 Having a VAT has also been associated with greater use of evidence-based CLABSI prevention practices,7 and low complication rates.8 9 Yet, our results also demonstrate the important role these teams play across the continuum of care for patients with specialised vascular access needs. This includes guiding device use in accordance with appropriateness criteria; preventing care delays by being more readily available when PICC placement is required; and, managing device-related complications, such as occlusions, that can result in treatment interruptions or premature device removal. Finally, VATs appear to be an integral part of preparing patients to perform critical self-care tasks postdischarge. These findings add to current knowledge by highlighting the multiple roles of VATs as a hospital resource and provide the elements for a business case for team investment given the potential savings from preventing harm and improving device use.
PICC use begins with the vascular device decision, which requires considering factors related to complication risk, including the required therapy, infusion duration, device type and patient factors.1 Our findings indicate that sites with VATs had generally well-defined decision-making processes. This included subspecialist consultations to determine therapy type and/or if a PICC was appropriate based on patient characteristics such as presence of chronic kidney disease. However, subspecialist involvement in decision-making was more variable or even lacking at the non-VAT sites. Studies suggest that PICCs are inappropriately used in patients with renal dysfunction24 or for patients who may not need a PICC for parenteral antimicrobial therapy postdischarge.25 Thus, interventions that promote and facilitate subspecialist consultation prior to PICC insertion could be important, especially at sites without VATs, to reduce potential hazards.26
VAT sites also used explicit decision-making criteria, namely MAGIC,1 which emerging evidence suggests is associated with improved safety and lower overall PICC use.5 27 28 Nonetheless, many ordering clinicians were not aware that appropriateness criteria existed, and no formal criteria were used at the non-VAT sites. This finding suggests a universal need for better education and strategies to seamlessly facilitate decision-making, such as embedding appropriateness criteria as part of a PICC order set, as was done at one VAT site. However, implementing appropriateness criteria also requires that alternative devices, such as midlines or ultra-sound guided peripheral intravenous catheters, are available, which was not the case at all sites. Additionally, alternative use at some sites was limited by clinician lack of awareness, or perceived lack of usefulness. Increasing awareness, availability and use of alternatives could therefore be another strategy for improving patient safety.29 30
A notable difference in PICC insertion processes, especially between sites with and without VATs, was the potential for care delays. Such delays were most striking at the site without on-site insertion capability, as patients had to be transported to another facility for PICC insertion. However, potential delays were also described at sites with primarily IR insertion due to lack of 24/7 IR clinician availability and PICC insertion being just one of many IR procedures. Concerns about delays were less common at VAT sites. Yet, one hospital was reducing VAT hours based on productivity (ie, number of devices placed within a certain time frame) and concerns about nursing staff maintaining proficiency with inserting peripheral intravenous catheters, while the other was increasing team availability by broadening the range of dedicated responsibilities to include responding to in-hospital cardiac arrests and providing nearly full-time vascular device management across the facility. These contrasting cases suggest a need for further research to assist hospitals in identifying strategies to capture return on investment related to developing and deploying VATs.
In-hospital PICC care also differed across sites. At all sites, RNs only were authorised to provide PICC care, although task delegation, scope of authorised practice and level of competency varied. At hospitals with VATs, the team RNs were responsible for many aspects of PICC-related care, including daily line checks, line malfunction management, and removal. However, bedside RNs were also part of the process, conducting regular assessments, blood draws, and as-needed dressing changes. At one of the hospitals without a VAT, a dedicated intravenous team functioned in much the same way. These models ensure that individuals with specialised knowledge oversee and perform critical management tasks to ensure optimal device care and function. On the other hand, maintaining competency and comfort with PICC care among bedside RN staff is also critical, given limited or, in some cases, no availability of these highly specialised providers. Thus, strategies to ensure competency are needed, especially at sites with low-volume PICC use. This could include regular competency assessments, as well as developing linkages between sites with and without VATs for consultation and ongoing education and training.
Home infusion therapy relies on patients and/or their informal caregivers performing vascular device self-care,15 16 tasks often learnt just prior to hospital discharge. Our study suggests there are no clear standards for teaching patients about PICC care, as well as variation in discharge processes and postdischarge care across hospitals. Hospitals with VATs had specific discharge teaching approaches and materials, including skills-based training (eg, device flushing). However, the process appeared haphazard at non-VAT sites, including lack of clarity about who had primary responsibility for PICC-related discharge education and educational content. While some of these shortcomings might be addressed by home care providers, the skill level and availability of these providers also vary.31 32 Given the importance of PICC-related self-care tasks in the delivery of safe, effective home infusion treatment, these gaps in discharge education and postdischarge care represent a critical area for further research, as well as a clear opportunity for care improvement. This includes developing discharge-specific interventions, as well as ways to support care outside the hospital, which could involve the use of virtual visits or technology.33 34
Our study has limitations. We conducted site visits and interviews with healthcare personnel at five purposefully selected VA Medical Centres. While these medical centres were selected in accordance with the predominant types of PICC insertion models based on prior research,7 this selected sample is not intended to be broadly generalisable or even representative of all VA Medical Centres or US acute care hospitals. Another limitation is the potential for response bias. To minimise potential bias, we interviewed individuals with a diverse set of organisational roles and thus varying perspectives and experiences related to PICC use at their facility. We also need to acknowledge that this work is largely descriptive and does not provide a deep analysis of the underlying factors driving the observed variation across hospitals. Finally, our interpretation of the findings and recommendations for improvement may reflect certain underlying biases based on the research team composition and perspectives. To mitigate this risk, our team was highly disciplinary and comprised two physicians, one nurse and four non-clinicians.
Our qualitative study identifies key aspects involving decisions about PICC use, the insertion process, PICC management and patient discharge education that benefit from the presence of a VAT. We found notable practice variations across hospitals and specific areas for both care improvement interventions and further research. This includes developing and evaluating strategies that improve clinician awareness and use of PICC appropriateness criteria and alternatives, informing the efficient and effective deployment of dedicated VATs, and addressing deficits in the patient discharge education process. Moreover, this research indicates that ongoing attention and rigorous practices across all the key process domains remain essential for ensuring appropriate, safe and effective PICC use.
Data availability statement
Data are available upon reasonable request. Deidentified, anonymised data that support the findings of this study will be made available by the corresponding author, SLK, on reasonable request and with a letter of agreement.
Patient consent for publication
This study was approved by the VA Ann Arbor Healthcare System institutional review board (IRB-2017–1038), with written documentation of informed consent obtained from interview participants by a member of the interview team.
The authors thank all the clinical staff who participated in interviews and the individuals at each site who assisted with site visit and interview logistics. In particular, we thank the site point of contacts who played an instrumental role in our visits to each of the hospitals.
Twitter @Sarahlkrein, @vineet_chopra
Contributors SLK and VC: study concept and design, conduct interviews, analysis and interpretation of data, drafting of the manuscript and critical revision of the manuscript for important intellectual content. MH: study design, conduct interviews, analysis and interpretation of data, and critical revision of the manuscript for important intellectual content. MQ and KEF: conduct interviews, analysis and interpretation of data, and critical revision of the manuscript for important intellectual content.LEW and BRG: analysis and interpretation of data, and critical revision of the manuscript for important intellectual content.
Funding This study was funded by US Department of Veterans Affairs (HSR&D IIR 15-313 and RCS 11-222). The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Department of Veterans Affairs or the US Federal Government.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.