Case Management in Primary Care for Frequent Users of Health Care Services: A Mixed Methods Study

Design

We used a sequential explanatory mixed methods design, with consecutive quantitative and qualitative phases.¹⁵ The first phase was a pragmatic randomized controlled trial with delayed intervention for the control group and measurements before and after the 6-month intervention.¹⁶ This phase of the study is reported according to the Consolidated Standards of Reporting Trials (CONSORT) guidelines.¹⁷ The second phase used a descriptive qualitative approach to gather a more comprehensive understanding of participants’ experience with the case management intervention.¹⁸ This study was approved by the research ethics board of the Centre intégré universitaire de santé et de services sociaux du Saguenay-Lac-Saint-Jean, Quebec, Canada.

Recruitment and Randomization

Within a pragmatic context, we identified frequent users by combining metrics on frequency of use and health care professionals’ judgment, according to current recommendations.¹⁹ All family physicians in 4 participating clinics received a list (generated by MAGIC Chronique software, MédiaMed Technologies²⁰) of the top hospital services users. We defined frequent users as patients with 3 or more emergency department visits, hospitalizations, or some combination thereof in the previous 12 months. Family physicians then identified patients with complex care needs who they felt could benefit from the case management intervention and suggested additional patients who were frequent users of their clinic. Eligibility criteria required that patients be aged 18 to 80 years and have at least 1 chronic disease (diabetes, cardiovascular disease, respiratory disease, musculoskeletal disease, or chronic pain). Patients with serious cognitive problems were excluded.

Six nurses received 50 hours of theoretical and practical training on case management, including self-management support techniques. The case management nurses contacted patients by telephone to ask if they were interested in participating. If a patient expressed interest, a research assistant telephoned the patient to explain the project, assess eligibility, and obtain consent. Participating patients were randomized into either the control or the intervention group according to a 1:1 ratio and based on a 3-stage randomization process: (1) generation of a random sequence using simple randomization, (2) provision of group allocations within sealed opaque envelopes, and (3) group allocation after patients’ enrollment in the study. The investigators who collected the data were blinded to the assignation and were not involved in the randomization process.

Intervention and Control Groups

The intervention group received the V1SAGES case management intervention for 6 months. The intervention had 4 required components: (1) evaluation of patient needs and resources, (2) implementation and sustainability of an individualized service plan tailored to patients’ priorities with the collaboration of the health care and community partners, (3) care coordination among the health care and community partners, and (4) provision of self-management support for patients and their families (eg, motivational interviewing). ^14,21–25 Each nurse had a maximum caseload of 50 patients.

Throughout the intervention, the case management nurse reassessed patients’ needs and progress. Frequency of the contact depended on a patient’s need. For the most part, these meetings were face to face, although the nurse was available to patients by telephone. Following the Stanford Chronic Disease Self-Management program, the patients and their spouses were also invited to 6 weekly 2.5-hour group meetings led by peer leaders. Meetings covered (1) techniques to deal with issues such as frustration, fatigue, pain, and isolation; (2) appropriate exercise for maintaining and improving strength, flexibility, and endurance; (3) appropriate use of medications; (4) effective communication with family, friends, and health professionals; (5) nutrition; (6) decision making; and (7) how to evaluate new treatments.

Any service that was provided to the patient in the intervention was reported to, or coordinated with, their family physician, and documented in their medical record. The case management nurse acted as the main point of contact and was informed of the patient’s emergency department visits and hospitalizations during the intervention period.

Patients in the control group received the usual care from primary care nurses (not trained in case management) for 6 months (health promotion, global evaluation, follow-up of chronic diseases). Then, after data collection at that time, they received the V1SAGES 6-month case management intervention.²⁶

Quantitative Data Collection and Analysis

Patient recruitment occurred from February 2013 to January 2014. At baseline, the research assistant administered the questionnaire to all patients at their primary care clinic and provided assistance as needed. Six months later, the research assistant collected outcomes data by telephone.

We assessed 2 outcomes: psychological distress measured with the Psychological Distress Scale (K6)²⁷ and patient activation measured with the Patient Activation Measure (PAM).^28,29 Psychological distress was analyzed as a dichotomous variable where scores of 13 or higher indicated the presence of psychological distress; patient activation was analyzed as a continuous variable. We also collected covariables at baseline to describe patient characteristics: (1) socioeconomic data (age, sex, marital status, education, occupation, economic status with family income, and perception of his/her economic situation); (2) health literacy, measured with the Newest Vital Sign (NVS)³⁰; (3) mental health, measured with the Hospital Anxiety and Depression Scale (HADS)^31,32; and (4) multimorbidity, measured with the Disease Burden Morbidity Assessment (DBMA).^33,34

Treatment fidelity was assessed to determine whether the case management intervention was delivered as intended.³⁵ For this purpose, data regarding delivery of the 4 main components of the intervention were collected from the medical records. Treatment fidelity was expressed as the proportion of case management intervention components delivered.

We sought to recruit all frequent users who were eligible and agreed to participate in the study, estimating an average of 50 to 75 frequent users per clinic.

All statistical analyses were performed according to the intent-to-treat principle. Missing values at 6 months were imputed from baseline. We first confirmed that there was no intracluster correlation (no nurse-specific effect) and that multilevel analyses were not necessary. We then described the characteristics of patients in each group at baseline, using means and standard deviations (for continuous variables) or percentages (for categorical variables). Groups were compared on presence of psychological distress at 6 months using logistic regression analysis, adjusting for baseline presence of psychological distress. No other variable was included in this analysis because there were no stratification variables to take into account.³⁶ Similarly, for patient activation, we compared group means at 6 months, adjusting for baseline scores, using analysis of covariance (ANCOVA). All statistical tests were performed with significance at the 5% level.

Qualitative Data Collection and Analysis

We contacted a sample of 69 key informants involved in the quantitative phase of this study to participate in individual semistructured interviews or focus groups after the 6-month trial. A maximum variation sampling³⁷ based on categories of participants, age, sex, and affiliated primary care clinic was used to capture a wide range of perspectives. In-depth semistructured individual interviews were conducted with 25 patients from the intervention group, 6 case management nurses, and 9 health managers. We also completed 3 focus groups with 8 patients’ spouses and 4 focus groups with 21 family physicians. The 1-hour interviews were conducted by an author trained in qualitative research (D.B.). The interview guide included open-ended questions and covered stakeholders’ perception of the case management intervention. Interviews and focus groups were continued until data saturation was achieved.³⁸ The dialog was digitally recorded and transcribed verbatim.

Data were analyzed by 2 authors from different backgrounds (F.D., a graduate in social work, and M.L., a graduate in anthropology) using mixed coding as described by Miles et al.³⁹ The reliability of the analysis was enhanced through data triangulation and peer debriefing during regular team meetings. NVivo 10 software (QSR International Pty Ltd) was used for qualitative analysis.

Integration of Quantitative and Qualitative Data and Analyses

The quantitative and qualitative findings were compared and contrasted through discussion among the authors. We considered and discussed convergent as well as divergent findings to achieve a more complete understanding of the effect of the case management intervention.¹⁵

Quantitative Results

Figure 1 shows the flow of patients through the trial. A total of 247 patients were recruited and completed the questionnaire at baseline; 126 patients were allocated to the intervention group, of whom 111 (88.1%) completed the intervention. The intervention and control groups had similar baseline characteristics, as shown in Table 1. Overall, 230 (93.1%) of the patients had follow-up through the study and completed the questionnaire at 6 months.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow of patients through the randomized trial.

ED = emergency department.

Case management nurses delivered each component of the intervention to patients as intended, demonstrating good intervention fidelity: 98.3% for patient evaluation, 94.8% for individualized care plan, 72.6% for care coordination, and 86.5% for self-management support.

Table 2 shows changes in prevalence of psychological distress and in level of patient activation. Compared with usual care, the intervention reduced the likelihood of psychological distress (odds ratio = 0.43; 95% CI, 0.19–0.95, P = .04) but did not significantly alter patient activation (Cohen d = 0.17; P = .43).

Qualitative Results

Characteristics of participants in the study’s qualitative phase are presented in Supplemental Tables 1 and 2 (available at http://www.annfammed.org/content/16/3/232/suppl/DC1/). The overall perception of stakeholders regarding the case management intervention was positive. Many of them noticed that improved accessibility and self-management support gave patients and spouses a sense of security and helped patients to better self-manage their health, as discussed in more detail below.

Findings in Context

The sense of security described by most of the interviewed patients and their spouses is consistent with the reduction in psychological distress observed in the pragmatic randomized controlled trial phase of this study. This finding is novel, and considering that frequent users tend to have higher psychological distress, ^11–13 this case management intervention is a promising avenue to enhance patients’ sense of security by facilitating access to care, providing a single point of access, and developing close relationships between the patient and the case management nurse.

Our findings are consistent with results from previous studies of case management interventions for frequent users. A qualitative study by Grinberg et al⁴¹ on the key components of successful case management reported that the development of a healing relationship between clinicians and their patients was based on security, genuineness, and continuity of care. A nonrandomized but controlled study by Adam et al⁴² demonstrated an improvement in patient well-being after the implementation of a case management intervention by an interdisciplinary care team. In addition, the evaluation of a community matron service by Leighton et al⁴³ found that a better access to health care services was related to patient confidence.

Participants in the qualitative phase of our study described improvement in patient self-management capacity. The qualitative study by Grinberg et al⁴¹ similarly suggested that the healing relationship between clinicians and their patients, based on security, could motivate the patient to engage in self-management. Another study, by Chapman et al,⁴⁴ with 31 health care professionals demonstrated that case management met patient needs through patient education about self-management of their conditions.

The improvement in self-management described in these studies was not seen in our pragmatic randomized controlled trial, however. This difference could be partly explained by the fact that we used an instrument to measure activation, an outcome that has been shown to be important in the care of patients with chronic conditions,⁴⁵ but not directly self-management. The PAM may have failed to capture changes in self-management; thus, a more specific tool to measure this outcome (eg, the Partners in Health scale⁴⁶) should be used in future studies. It is also possible that a 6-month follow-up period is not sufficient to observe a significant change in frequent users’ self-management.⁶

Considering that 30% of intervention group patients had depression or anxiety symptoms at baseline, it is possible that it took a few months for case management to reduce their anxiety so that they could be receptive to self-management support. Indeed, patients with higher psychological distress could be less receptive to such support.⁴⁷ Stakeholders perhaps began to note positive effect of the intervention on self-management for certain patients with less psychological distress or after reduction of their psychological distress, but more time would have been needed to capture this improvement in the trial.

Strengths and Limitations

Mixed methods designs provide a deep understanding of a complex intervention and its multidimensional factors. They use the strength of each method to get a more complete and accurate picture of the phenomenon. ⁴⁸ The quantitative phase of this study was strong in that pragmatic randomized controlled trials produce more reliable evidence compared with before-and-after studies. The qualitative phase benefited from the varied backgrounds of the coauthors (family medicine, nursing, social work, and anthropology), which allowed for a rich triangulation of perspectives when analyzing the qualitative data. Moreover, the sample size of 69 participants was ample to reach data saturation. We used a recommended case-finding procedure (hospital data and health care professionals’ opinions) to identify patients to participate in our case management intervention.¹⁹

At the same time, several limitations should be considered. During the case management intervention, 24 patients in the intervention group also participated in the Chronic Disease Self-Management Program. Although this is a small proportion of the control group, it could have introduced a cointervention bias. The 6-month evaluation period was too short to fully implement and evaluate such a complex intervention. Evaluating effect on services use and cost-effectiveness would also have been interesting, but would have required a larger sample size. In addition, although the case management nurses had no contact with the control group, they may have unintentionally influenced other nurses who may have adopted a case management approach with the patients they cared for in the control group.

Conclusions

In conclusion, our findings suggest that the V1SAGES case management intervention reduces psychological distress, giving patients a greater sense of security. Many patients and stakeholders reported improvement in self-management capacity, but this impression did not translate into improved patient activation in the pragmatic randomized controlled trial. Case management holds promise for improving patient outcomes. Future studies could evaluate effect of the case management intervention on service use and costs, and also assess whether a longer intervention duration (1 year) would improve self-management.