Integrating Primary Care Into Assertive Community Treatment

This study included 378 participants from six ACT teams associated with a New York City behavioral health agency from January to June 2019. Available participant demographic characteristics were as follows: 132 (46%) participants were female, and 155 (54%) male; average age was 44 years (range 21–82); 178 (62%) identified as Black, 52 (18%) as Hispanic, 37 (13%) as White, and 20 (7%) as other or not identified.

Five ACT teams in Brooklyn were assigned primary care nurse practitioners from a FQHC-based postgraduate fellowship training program 2 days per month during the study period. Four of the ACT teams were colocated with the FQHC, and two of the teams were newer and “shelter based” (one team was both coloated and shelter based), with referrals exclusively from a few homeless shelters assigned to each team. The five ACT teams assigned a primary care nurse practitioner had a total of 305 participants. (Characteristics of the ACT teams and the degree of primary care support are described in an online supplement to this report.) One ACT team in the Bronx served as a control, with no assigned primary care nurse practitioner.

On the days they spent with the ACT team, the primary care nurse practitioners attended ACT team meetings to discuss participants’ medical concerns and attended field visits with an ACT team member to geographical clusters of participants that included two to four participants who had unmet medical needs. The primary care nurse practitioners working with the FQHC-colocated teams also attended the weekly ACT wellness group and luncheon, during which they had opportunities to engage participants, including during walk-in visits to the FQHC when appropriate. The ACT psychiatrists also consulted the primary care nurse practitioners regarding medical issues as needed, often outside of their scheduled ACT days. The primary care nurse practitioners were supervised by more experienced primary care preceptors at the FQHC. Because of onboarding and scheduling issues, two of the primary care nurse practitioners spent only 2 days each with their teams in the 6 months.

The data for this pilot study were obtained from focus groups and a cardiometabolic data registry we developed. Six focus groups were conducted by two external evaluators with the following convenience samples: staff from each of the five Brooklyn ACT teams during their morning meetings (N=20 staff members) and 16 ACT participants attending a weekly wellness group. One additional focus group was conducted by a third external evaluator with the five primary care nurse practitioners. The evaluators posed the following questions in all focus groups and recorded responses with concurrent note taking: What was it like to have or be a family (primary care) nurse practitioner on the team? Did having a family nurse practitioner on the team benefit participants? In what ways? How has staff’s work changed with the addition of a family nurse practitioner? What suggestions would you give an ACT team getting primary care (on their team) for the first time? Is there anything else we missed that you would like to add? The evaluators reviewed focus group notes and identified emergent themes. Where two evaluators were present, they compared notes and themes for alignment.

A database of cardiometabolic risk factors, including weight, blood pressure, hemoglobin A1c, and cholesterol, was created for each team. At baseline (January 2019) each team populated the spreadsheet with all data obtained in the preceding year for the current ACT participants. New measurements collected were entered in the ensuing 6 months. The proportion of participants for each team who had hemoglobin A1c and cholesterol screenings performed in the preceding year postintervention (July 2019) was compared with the proportion at baseline (January 2019), and chi-square tests were used to test for significance. Binary ordinal logistic regression was used to determine whether changes over time were significantly different between the established ACT teams and the newer, shelter-based ACT teams. This pilot study was determined by the agency’s institutional review board to be exempt research.

From January to July 2019, 69 of the 378 participants on the six ACT teams were seen at the FQHC, for a total of 263 primary care visits, of which 66 were walk-in visits. A count of field visits and consultations made by the primary care nurse practitioners was not maintained.

The evaluators who conducted the focus groups reviewed the data and identified three themes. The first theme was that the addition of primary care to ACT services supported ACT participants’ ability to trust the primary care nurse practitioners. Staff described numerous examples of people they had been attempting to engage in primary care for months to years with little success. The ACT teams were able to successfully build trust with individuals over time, but this trust did not necessarily extend to other providers. This was true even with ACT teams operating in shelters with onsite medical clinics. However, when the primary care nurse practitioners were introduced as part of the ACT teams, these individuals were willing to agree to a “warm handoff” (in-person introduction) to the FQHC. For some individuals, an opportunity for a visit would be lost if not acted on at the moment that the individual expressed interest or acquiesced to a visit. Trust in the ACT team was the motivating factor for connecting to care. Participants in the ACT teams reported that the primary care nurse practitioners were more accommodating of people with psychiatric diagnoses, and they felt more understood and respected, compared with past experiences. When participants were asked about the benefits of being connected to primary care, responses included that it helped them feel less anxious, prevented long waits in the emergency department, and enabled them to see one provider consistently, allowing the primary care nurse practitioner to get to know them and become a trusted provider. In addition, ACT participants attending the wellness group reported that it was convenient to have a primary care clinic located in the same building as the ACT office.

The second theme that the evaluators identified was that the availability of the primary care nurse practitioners helped both participants and providers to access health information. Staff on the ACT teams observed that participants, once engaged, seemed to respond more positively to discussing health conditions with the primary care nurse practitioners than with the ACT team, viewing the primary care nurse practitioners as more of an expert in this area. The primary care nurse practitioners were able to answer detailed questions about medical conditions and medications, which alleviated anxiety for ACT participants and their families. The ACT nurses in particular emphasized the advantages of the primary care nurse practitioners’ ability to prescribe medications on the spot to better manage chronic health conditions in collaboration with the psychiatrists. The psychiatrists found ad hoc consultation with the primary care nurse practitioners, often through e-mail, phone, videoconferencing, or text, to be one of the more helpful features of the integration.

The third theme identified from focus groups was that primary care field visits did not appear to be the most effective way of serving most ACT participants. The primary care nurse practitioners noted that guidance was limited because their preceptors lacked experience in field-based primary care. During field visits, there was uncertainty about which physical exam procedures could be performed, and sometimes the primary care nurse practitioners felt that the procedures they were able to do, such as measuring vital signs or performing phlebotomy, could be done by others. Phlebotomy in the field was cumbersome and performed only on two occasions, when they were required for homebound participants. Staff identified the difficulty of finding space for physical exams and procedures in shelter or street settings. They also reported that the role of the primary care nurse practitioners was unclear when nurses visited participants already engaged in other primary care services.

Analysis of data from the cardiometabolic registry revealed changes in screening rates over time. After exclusion of discharged participants, data were available for 274 individuals in January 2019 and 305 in July 2019. Over the 6-month pilot period, the proportion of ACT participants who had hemoglobin A1c screenings increased significantly from 12% (N=34) to 34% (N=104) (χ²=37.40, df=1, p<0.001) and cholesterol screenings from 16% (N=44) to 36% (N=109) (χ²=28.75, df=1, p<0.001) for the teams with integrated primary care nurse practitioners. (Results for specific teams are available in the online supplement.) No significant change was found in screening rates for the Bronx team that had no integrated primary care nurse practitioner; rates were 11% (N=35) in January 2019 and 7% (N=73) in July 2019. No significant differences in change in screening rates over time were found between the established teams and the newer shelter-based teams. For the ACT teams colocated with the FQHC, 85% (N=79) of participants who received screenings during the pilot period received them at the FQHC.

We constructed the following vignettes, based on our clinical experience during the pilot study, to illustrate aspects of the integration that may have improved client health.

Colocation facilitated warm handoffs for a participant in his 60s with schizoaffective disorder, diabetes, hypertension, obesity, and worsening vision who had not been able to make and keep a primary care appointment in 2 years because of psychosis and insurance issues. He was first engaged by a primary care nurse practitioner in the ACT office and eventually agreed to be seen in the clinic. After starting a long-acting diabetes medication, his glucose readings went from the 400s to 200s. This experience of success led to him taking increasing responsibility for his health, first by taking some recommended oral medications at home, then by decreasing his sugar intake. The primary care nurse practitioner later used an online application (RubiconMD) to request virtual consults from a cardiologist and neurologist after an episode of arrhythmia, obviating the need for specialty visits for this man who had a history of difficulty keeping appointments.

In another instance utilizing technology, the ACT psychiatrist was able to obtain instruction about the dosing and use of an insulin pen via videoconferencing to the primary care nurse practitioner for a participant in her 50s with schizophrenia and very poorly controlled diabetes. This participant was unable to return to the clinic because of psychotic disorganization and other stressors, so diabetes care was largely provided via ongoing teleconsultation between the primary care nurse practitioner and the psychiatrist.

A participant in his 80s with bipolar mania was discharged home by a nursing facility because of an inability to cooperate with rehabilitation treatment for pressure ulcers. Home wound care was not immediately available, and the participant was unable to attend follow-up appointments because of mobility issues. Access to a primary care nurse practitioner who could provide wound assessment and teach wound care helped the team to bridge care safely until external services became available.

In this pilot study, we found that ACT staff and participants accepted and valued integration of the primary care nurse practitioners because of the increased participant engagement and consultation-supported care, in some cases facilitated by colocation. Feedback regarding the importance of a warm handoff, whether in person or virtual, points to the primacy of trust in health care engagement.

Increased cardiometabolic screening rates over the 6-month pilot study for the five Brooklyn ACT teams suggested that integration of primary care nurse practitioners into the team improved participants’ access to primary care. The increase was not explained by the use of a data registry, which was also used by the comparison ACT team in the Bronx without an integrated primary care nurse practitioner, nor by the age of the ACT team or whether it primarily served shelter clients. The fact that the increase was found whether the ACT team had more (2 days per month) or less (2 days in 6 months) dedicated time from the primary care nurse practitioner might suggest that the team’s relationship with a primary care nurse practitioner was more important than the dedicated time for field visits. In future studies, researchers should examine whether cardiometabolic risk measures, adverse health events, and health care costs are affected by integration of a primary care provider.

The heterogeneity of the teams, including in how primary care was integrated, was a limitation of this study. However, the opportunity to pilot variations of the integration model highlighted some interventions that were more helpful than others and laid the groundwork for further studies. Colocation of the FQHC, and its effect not only on logistics but also on the relationship between primary and behavioral health clinicians, was an important factor supporting integration. The ACT participant focus group may have been biased by selection from the wellness group held in the same building as the FQHC, but the participants’ preference for colocation of the FQHC was clear. For other participants, teleconsultation, made possible by the partnership fostered between the ACT teams and the FQHC, transcended physical barriers to integration.

Feasibility issues were encountered when using nurse practitioner fellows as the primary care providers and when the focus was on field-based visits. Our intervention involved nurse practitioner fellows in order to minimize cost and maximize sustainability, while promoting education among primary care providers about the care of people with serious mental illness, but it was a challenge for the newly graduated primary care nurse practitioners to provide primary care support in a complex environment. One strategy may be to pair the primary care nurse practitioner trainee with a preceptor experienced in home care or street medicine, using the model of the Department of Veterans Affairs’ Home Based Primary Care program (8). Nonetheless, our findings suggest that primary care nurse practitioners are best employed to facilitate engagement with the FQHC, because screenings were most often successfully obtained in this way in the pilot study.

Another path to sustainable primary care integration may involve leveraging a psychiatrist’s ability to provide some primary care, including management of common conditions such as hypertension, diabetes, and obesity. This approach would be particularly helpful for participants who are not engaged with primary care. Psychiatrists in ACT teams could treat common conditions after a careful evaluation of urgency and complexity of the condition or conditions, participants’ access to primary care, the psychiatrists’ knowledge base and comfort level, systems support, and participant preference (9). One integration model that uses a consultant to efficiently bring a scarce service to a large population is IMPACT (Improving Mood Promoting Access to Collaborative Treatment), where a psychiatrist consultant supports depression care for a primary care population. This is achieved by means of weekly supervision of a depression care manager who serves as a liaison for primary care providers, data registry of routine depression screenings to monitor patient progress and risk, and consultation to primary care providers on psychopharmacology (10).

A mirror image of the IMPACT model could involve a primary care provider meeting weekly with ACT team members to provide primary care recommendations for participants whose cardiometabolic medications are managed by the psychiatrist. Some flexibility for walk-in visits and possibly field visits may be necessary for participants whose engagement is more challenging. Using what we have learned from this pilot study, our group will be implementing and evaluating this proposed model.

Integrating primary care into the ACT model has the potential to improve health outcomes for people with serious mental illness, but the optimal frequency and type of primary care intervention still needs to be determined. A colocated primary care provider available for engagement of participants and consultation to the ACT team, utilizing a cardiometabolic registry to guide care, may offer a sustainable model of integration.

The authors thank Elisa Chow, Ph.D., Robert Hayes, J.D., Elizabeth Dubois, D.N.P., Colette Russen, F.N.P.-C., and Karin Olson, Ph.D., for their work on this project.