Lessons Learned From Implementing a Caring Contacts Clinical Practice Guideline Recommendation

During a 1-year period, veterans hospitalized in a psychiatric inpatient unit of a large VA health care system were identified for potential enrollment. Each Tuesday and Friday, a caring contacts program coordinator used electronic health records (EHRs) to screen for eligible veterans. A program coordinator position was established because previous studies have shown that the challenges of tracking hundreds of individual mailing schedules, avoiding repeat enrollments (i.e., after repeated hospitalizations), and similar administrative burdens should not be underestimated (3). The primary screening was conducted twice weekly on the basis of the available resources (i.e., the program coordinator was not available every day) and predictions that few veterans would be missed (based on the average length of stay in this unit). Veterans were excluded if they had a behavioral flag in the EHR that suggested that caring contacts may be contraindicated (e.g., boundary problems with clinicians) or if they were already enrolled in the Recovery Engagement and Coordination for Health–Veterans Enhanced Treatment (REACH VET) caring contacts program (3). Participants in REACH VET were excluded because the cards for the newly implemented inpatient caring contacts program were developed on the basis of the REACH VET model, so they were very similar. Veterans enrolled in other caring contacts programs (described below) were not excluded because some of these programs are required by policy and the nature of the cards sent by the programs was quite different (5).

An inpatient clinician conducted a secondary screening where veterans were deemed ineligible if they had an invalid mailing address (e.g., were homeless) or if the intervention was not clinically appropriate. Reasons for ineligibility were recorded. Eligible veterans were approached by the clinician who described the program and asked the veterans whether they would like to opt in.

Although we were interested in traditional QIP outcomes (i.e., acceptability and barriers), we also identified overlap with other VA programs as a key metric for implementation. The VA has three other optional or required caring contacts programs for veterans at high risk for suicide. One is based on the VA’s use of high-risk flags in the EHR (6). A second is for callers to the Veterans Crisis Line (VCL) (5). Finally, clinicians are encouraged to consider caring contacts also for veterans identified by the REACH VET program (7). Therefore, implementation of the CPG recommendation raised the possibility that a veteran could receive caring contacts from four different programs. The optimal “dose” of caring contacts is currently unknown.

During the first year of the program, 462 unique veterans were hospitalized on the unit, and 418 (90%) were considered for enrollment in the caring contacts programs (because the primary screening occurred only 2 days per week and some veterans were missed). After the primary screening by the program coordinator, 90 veterans (22% of all veterans considered) were not further assessed for eligibility because of the unavailability of inpatient providers to conduct the secondary screening. Of those who were fully screened, 163 veterans (39% of all veterans considered) were ineligible for the intervention; the most common reasons for ineligibility were homelessness or housing instability (19%, N=61 of 328), enrollment in the REACH VET caring contacts program (13%, N=43 of 328), and behavioral flags (7%, N=23 of 328). Ultimately, 135 veterans (of 165 invited, 82%) opted in and were enrolled. As of March 4, 2022, 833 letters were sent to enrolled veterans. Caring contacts were discontinued for 13 veterans (10% of those enrolled); the most common reason for discontinuation was an invalid address (39%, N=5).

QI methods often result in opportunities for data-driven procedural refinements in clinical settings such as a busy psychiatric inpatient unit. Because only 29% (N=135 of 462) of hospitalized veterans were enrolled, we note opportunities to improve reach in the next phase of the development of the caring contacts program. Consistent with the plan, do, check, act cycle, potential improvements that could be tested in additional brief QI cycles include the following.

First, the primary screen by the coordinator was intended to save time for busy clinicians, but it limited the enrollment window to 2 days per week (so about 10% of veterans were missed). Additional resources could be used to increase enrollment days, or the primary screen could be eliminated.

Second, 22% (N=90) of all veterans considered for the intervention were lost because inpatient providers were too busy for the secondary screen. Other resources could also be added to the program so that inpatient providers are not needed to conduct the enrollment procedures. Alternatively, steps could be incorporated into routine business processes (e.g., use of a checklist during morning rounds).

Third, most caring contacts interventions do not require patients to opt in, and the preliminary QI data indicated very high acceptance of the intervention among veterans; the caring contacts cards could therefore be sent to all eligible veterans.

Fourth, exclusion criteria could be simplified. For example, the content of the intervention could be changed so that it would not be very similar to REACH VET. Veterans identified for REACH VET could then be enrolled also in the new inpatient program, especially if evidence accumulates to indicate that receiving multiple friendly letters does not dilute their effectiveness.

Fifth, additional problem solving may improve reach. For example, an address is required for this intervention, and many homeless veterans have only temporary addresses. Our inpatient unit may be able to partner with VA homelessness programs to obtain shelter addresses or similar temporary addresses. We did find that problem solving any issues with addresses was helpful; 13 caring contacts letters were returned to our hospital, of which 10 were resent after the program coordinator identified a new address. Alternative delivery methods could also be considered for homeless veterans in the future (e.g., text messages).

Initially, when the program launched, each veteran who opted into the program was asked to identify a preferred inpatient clinician from whom they would like to receive caring contacts. After 17 veterans were enrolled, 14 stated that they did not have any preference. For the three veterans who named a preferred clinician, all the clinicians worked in other areas of the hospital (i.e., did not work on the inpatient unit). In addition, three of nine clinicians from the unit preferred a generic signatory that referred to the unit staff. Furthermore, only two of the nine clinician signatories agreed to listing their direct phone number on the cards (rather than a general clinical contact number). On the basis of these results, a general signatory (7 West Care Team) and a general contact number were used for all cards.

In total, 73% of veterans who enrolled in the inpatient program were also enrolled in at least one other program (Table 1). Specifically, 68 veterans (50%) were also enrolled in the VCL caring contacts program, which had the highest degree of enrollment overlap among the programs.

The potential advantages and disadvantages associated with delivering multiple versions of caring contacts are unknown. Receiving numerous cards could diminish the intervention’s effect or could have counterproductive effects on the mechanisms of interest (e.g., caring connections). We examined 31 veterans who would have received at least one VCL card by the time they were asked to consider the inpatient caring contacts program; 28 of 31 veterans (90%) who presumably already received at least one VCL card still elected to opt into the inpatient program.

One advantage of overlapping caring contacts programs was that veterans could be reconsidered for enrollment in an evidence-based caring contacts practice. The REACH VET caring contacts program enrolled 48 veterans previously considered for the inpatient program who were never enrolled (e.g., because of incomplete secondary screenings or ineligibility). Among these 48 veterans were 15 who had been excluded because of homelessness at the time of the inpatient program screening and 24 with incomplete screenings. Ten veterans had declined the inpatient program but were later enrolled in REACH VET caring contacts. Overall, 26 veterans were disenrolled from the VCL caring contacts program because of an invalid address, of whom six were later enrolled in the inpatient caring contacts program.