Reasons for Patient Nonattendance During Acute Partial Hospitalization

Psychiatric partial hospital programs (PHPs) have been defined as “time-limited ambulatory active treatment programs that offer therapeutically intensive, coordinated, and structured clinical services within a stable milieu” (1). Along with other services, such as intensive outpatient programs and assertive community treatment, PHPs have been used as an alternative to inpatient hospitalization and as a step-down from inpatient care. Over 50 years, many studies have documented that PHP treatment can often substitute for inpatient care, may produce better outcomes, may be preferred by patients, and is usually less costly (2,3). Under the Patient Protection and Affordable Care Act, use of PHPs and related ambulatory treatments for acutely ill patients is likely to increase, as population-based health systems emphasize less expensive modalities and impose financial penalties for hospital readmissions (4).

Yet despite potential advantages, PHP also presents clinical and administrative challenges. Among these challenges is the problem of patient nonattendance. Partial hospital patients are outpatients, albeit with high levels of symptomatology, and they continue to live in the community during the admission. In contrast to inpatient admission, as one patient put it, “life just does go on even when I’m in the partial hospital.” Unlike inpatients, therefore, partial hospital patients must choose whether to attend the program on any given day in the face of competing responsibilities, practical barriers, and ongoing symptoms. This arrangement may pose clinical problems because patients who choose not to attend the PHP are deprived of structure, support, and treatment and may be at greater risk of decompensation. Nonattendance may also limit access by other patients awaiting admission and reduce program profitability.

The literature on PHP nonattendance is small and dates to an era in which lengths of stay in PHPs were much longer (5,6). Studies from that time showed that nonattendance occurred frequently and was associated with various clinical and demographic parameters, including substance abuse, multiple prior admissions, competing life responsibilities, and “negative” reactions to treatment. The relevance of this literature to current practice is uncertain, however. To our knowledge, no published study has explored the causes or consequences of nonattendance in PHPs as they are currently structured, typically as freestanding, acute care programs with brief lengths of stay and high patient acuity.

The purpose of this study was to examine patients’ nonattendance at an acute PHP. The following questions were addressed: What percentage of patients do not attend one or more days during their PHP admission? What are the clinical and demographic correlates of nonattendance? What reasons can be identified for patients’ nonattendance? Are patients who miss days in PHP more likely than those who attend daily to drop out or to require inpatient or PHP readmission?

Methods

The study was conducted between March and July 2013 in an acute PHP at Butler Hospital, a freestanding, not-for-profit psychiatric hospital in Providence, Rhode Island, associated with the Warren Alpert Medical School of Brown University. The study was approved by the Butler Hospital Institutional Review Board.

During the time of this study, the average daily census of the PHP was 47 patients and the mean±SD length of stay was 5.6±3.1 days. All private or public insurance was accepted, except Medicaid. All patients admitted from March 13, 2013, through July 13, 2013, were included in this study.

The program operates from 9 a.m. to 3 p.m., five days a week, and offers 4.5 hours of group programming daily for each patient. Programming comprises four therapy or psychoeducation groups based upon principles of cognitive-behavioral therapy (CBT) as well as brief “goal setting” and “check-out” groups. A component of the program, called “CBT-i,” employs more behavioral and less cognitively demanding techniques, such as group exercises or craft projects, for more impaired patients. A second, contiguous component is a dialectical behavioral therapy program for women with borderline personality disorder and related conditions.

In addition to participating in group programs, patients are seen individually each day by an attending psychiatrist for 20- to 30-minute psychotherapy or medication management visits as well as by other members of the clinical staff, as indicated. Whenever possible, with patients’ permission, family and outpatient providers are contacted and involved in the treatment. The PHP is staffed by ten full- or part-time psychiatrists (approximately seven full-time-equivalents), two Ph.D. psychologists, one advanced practice registered nurse, four registered nurses, 3.5 licensed independent clinical social workers, four activities therapists, one occupational therapist, and two occupational therapy assistants, in addition to residents and other trainees.

DSM-IV diagnoses are made by patients’ attending psychiatrists on the basis of clinical evaluation. Because of the many similar but nonidentical diagnoses that can be made clinically (for example, the many forms of depression), for these analyses, each patient’s primary diagnosis was assigned to one of six groups: depressive (major depression, dysthymia, and depression not otherwise specified), bipolar, anxiety, adjustment, psychotic, and other disorders. In addition, because patients commonly have more than one diagnosis and because this information is clinically important, any other additional diagnoses were also recorded and assigned to one of the six groups. The presence of a co-occurring substance use disorder was recorded separately. These diagnoses, as well as other clinical information (number of prior treatment episodes, number of kept and missed PHP days, and source of referral) and demographic data (age, gender, race-ethnicity, employment status including disability status, marital status, education level, insurance, and zip code of residence), were obtained from the medical record.

Clinical staff and a member of the research team monitored the program daily. Absent patients were always called if they did not attend daily treatment and were nearly always contacted on the same day. They were asked to provide a reason for not attending the program that day, and the information was reviewed with the attending psychiatrist.

For each day in which an enrolled patient did not attend the PHP—whether because of a planned absence for which he or she had been excused or an unplanned absence—the patient’s attending psychiatrist completed a one-page form indicating his or her understanding of why the patient did not attend and whether the absence was planned or unplanned. Ten possible reasons were provided: transportation issues, legal issues, inclement weather, acute general medical illness, a crucial medical appointment, child care, illness in the family, psychiatric illness including acute symptoms or side effects, change of psychiatrist, and other. More than one reason could be chosen.

To evaluate whether patients who did not attend daily treatment subsequently completed the program or dropped out, we tabulated the number of patients who were discharged on a day in which they were absent or the next day.

Results

During the four-month study period, there were 744 admissions to the PHP involving 675 patients. Admissions were predominantly females (N=515, 69%), with an average age of 40.1±14.7 years; 84% (N=625) were Caucasian, and 47% (N=352) were single. Forty-nine percent (N=364) had attended at least some college, but only 45% (N=338) were employed; 20% (N=150) were disabled. Most (N=646, 87%) had private insurance.

Eighty percent (N=598) had a primary diagnosis of an affective (depressive or bipolar) disorder, 63% (N=470) a co-occurring anxiety disorder, and 33% (N=243) a co-occurring substance use disorder (presently or in the past). Two-thirds (N=498) of patients’ episodes were admissions directly from outpatient status, whereas one-third (N=246) were transfers from one of the hospital’s inpatient units.

Patients had a total of 7.0±10.0 previous treatment episodes of any type at our institution (including outpatient treatment and visits only for evaluation) at this institution, including 2.2±5.1 inpatient admissions and 1.4±3.1 PHP admissions. Patients attended the PHP for 5.6±3.1 days. The CBT program, with 442 (59%) patients, was the largest component, and CBT-i, with 112 (15%) patients, was the smallest.

For 163 (22%) admissions, patients did not attend at least one day; for 78 admissions (10%), more than one day was missed. The average number of missed visits for patients who missed at least one visit was 1.1±1.4. Of a total of 4,496 patient-days covered by the study period, 294 (7%) were missed, most of which (N=260) were unplanned.

Nonattendance was correlated with several clinical and demographic characteristics, including number of prior inpatient episodes, total number of prior treatment episodes, total number of prior treatment episodes corrected for age, co-occurring substance use disorder, disability, fewer years of education, lower per capita income by zip code of residence, and attendance in a program for more impaired patients (Table 1). Several factors that were considered likely predictors of nonattendance were not associated with missing one or more days of treatment—they included primary DSM-IV axis I diagnosis, psychiatric diagnoses other than the primary diagnosis (apart from substance use disorders), source of admission (outpatient or inpatient), age, gender, race-ethnicity, marital status, employment status, insurance, and distance from home to the PHP.

Of the 163 admissions with an absence, 128 (79%) were related to acute general medical or psychiatric symptoms, including medication side effects, or a crucial medical appointment, according to the treating psychiatrists, and 89 (54%) were never completed, for an overall noncompletion rate of 12%. As the number of missed days increased, the chance of completing the program decreased. For example, if the admission was associated with only one day of missed treatment, 61% (N=48) of patients eventually completed the program, but if the admission was associated with four or more missed days, only 6% (N=1) of patients completed the program. The percentage of readmissions within 30 days was lower for admissions in which the program was completed versus admissions in which the program was not completed: readmissions accounted for 71 of 656 admissions of completers (11%) versus 18 of 95 (19%) admissions of noncompleters (χ²=5.24, df=2, p=.02).

Discussion

Nonattendance in outpatient psychiatric settings is common, and, as this study documented, it can also be a frequent problem during acute partial hospitalization, occurring for 22% of admissions. Although nonattendance may result from many clinical and practical factors, may deprive patients of important treatment components, such as structure and socialization, and may impair the functioning of PHPs, the causes and consequences of nonattendance during acute partial hospitalization have not been studied. In this PHP, factors associated with nonattendance also resembled those found in older studies of partial hospital programs and more recent studies of outpatient treatment. These factors included limited personal and economic resources, high rates of substance use disorders, multiple admissions, and disability (7–9). Unplanned absences and other treatment interruptions in other settings have been associated with negative outcomes, including readmission and mortality (10,11). This pattern also occurred in this study, which found that patients who missed one or more days tended to drop out and require readmission. Our findings thus extend previous results to partial hospitalization as it is currently provided.

Knowing correlates of nonattendance may help us understand why patients do not attend our PHP regularly and suggest options for improvement. The principal reasons identified for patient nonattendance in our PHP were general medical and psychiatric symptoms as well as other medical appointments—not “external” factors, such as transportation, child care, or weather. The fact that lower levels of education and income, having disability status, attending a program for more impaired patients, and a history of substance use disorders were associated with nonattendance suggest that patients with irregular attendance may have limited external and personal coping resources to handle ongoing symptoms or may feel uncomfortable and out of place among people who are more “advantaged.” Greater flexibility in program scheduling might help patients to balance treatment participation with other obligations, such as appointments, and allow them to attend when symptoms permit. Adjusting expectations to be more consistent with patients’ level of symptomatology and other clinical determinants of engagement, such as activation (12), might improve attendance and program completion. Greater attention to other relevant patient variables, including co-occurring substance use disorders, and to clinician characteristics, such as a capacity for empathy (13,14), may improve attendance and outcome.

This study had several limitations. This PHP is likely dissimilar to many other PHPs, having more physicians and a briefer length of stay. Given that our PHP is located in a freestanding psychiatric hospital, Medicaid patients were excluded. Yet in other respects, the program is not unusual: the time spent by patients in group each day (approximately 4.5 hours), the amount of individual therapy provided daily (approximately 20 minutes), and the focus on CBT are all typical of other daily partial hospital programs. In addition, we did not attempt to measure or infer patients’ attitudes toward the program, for example, whether they thought it was an appropriate fit, anxiety about participating in group settings, concern that problems were not being addressed, or a lack of connection with providers. Such factors are important determinants of treatment engagement and outcome and are likely important in PHP as well.

Conclusions

This study provides information on the causes and consequences of nonattendance of PHPs that may have practical implications in the current health care environment. Implementation of population-based health policies makes it especially important for providers to understand and treat patients who have difficulty participating actively because of a myriad of acute medical, psychiatric, and other social problems that happen to impact psychiatric outpatients. It becomes clear that acute partial hospitals must provide flexibilities to develop treatments more specifically targeted to such individuals.