Introduction 1 2 3 4 5 6 7 8 1 8 9 10 8 11 12 13 7 12 14 13 14 16 14 12 15 12 The purpose of the present study was to compare three groups of hospitalized psychiatric patients: (1) patients admitted and discharged voluntarily (VOL-VOL), (2) patients admitted involuntarily who later signed-in voluntarily and were discharged voluntarily (INV-VOL), and (3) patients admitted and discharged on an involuntary legal status (INV-INV). As such, this trichotomous criterion variable addressed the change (or lack thereof) in patients’ legal status during hospitalization, allowing an examination of whether or not certain sociodemographic and clinical factors are associated with differences in admission-to-discharge legal status. Treating legal status as a trichotomous variable (rather than solely considering either admission or discharge legal status) may yield meaningful results that add to the current body of research on correlates of patients’ legal status. Methods Subjects and setting 17 All patients were admitted to the 22-bed inpatient unit or the 8-bed crisis stabilization unit of a large, public-sector hospital in the southeastern United States. Both the longer-stay inpatient unit and the crisis stabilization unit admit patients primarily for evaluation and treatment of first episodes of illness or exacerbations of severe and persistent mental illnesses (primarily schizophrenia and other psychotic disorders and severe affective disorders). This county hospital serves a predominantly low-income, urban, African American population. 3 Procedures 17 18 Analysis Bivariate analyses were performed using a variety of independent variables in relation to the criterion variable. Continuous variables were analyzed using one-way analysis of variance and categorical variables were analyzed using chi-square tests of association. These bivariate analyses were conducted to inform the selection of variables for inclusion in the subsequent multivariate regression analysis. P P SAS version 9.1 Results 1 Table 1 n Variable Frequency Gender, female 126 (55.5%) Race     Black/African American 191 (84.9%)     Hispanic/Latino 4 (1.8%)     White/Caucasian/European American  28 (12.4%)     Other 2 (0.9%) Employment status, unemployed 213 (94.2%) Homeless on admission  58 (25.6%) Marital status     Single/never married 154 (67.8%)     Married/living as married  19 (8.4%)     Separated/divorced/widowed  54 (23.8%) Prior psychiatric hospitalization 169 (75.1%) Psychotic symptoms present on admission 194 (85.8%) Variable Mean ± SD Median Range Age  38.7 ± 12.4 40 17–76 Years of education  12.1 ± 2.3 12  4–18 GAF score on admission  30.3 ± 8.2 30 10–55 2 2 P 2 P Table 2 Sociodemographic characteristics of 227 hospitalized patients Variable a Test statistic (df) P n n n Gender, female 32 (57.1%) 48 (59.3%) 46 (51.1%) 1.23 (2) 0.54 Race, Black/African American 45 (81.8%) 66 (81.5%) 80 (89.9%) 2.87 (2) 0.24 Employed part-time or full-time 2 (3.64%) 7 (8.64%) 4 (4.44%) 1.99 (2) 0.37 Homeless on admission 17 (30.4%) 23 (28.4%) 18 (20.0%) 2.48 (2) 0.29 Current marital     Single/never married 32 (57.1%) 56 (69.1%) 66 (73.3%) 4.32 (4) 0.36     Married/living as married 6 (10.7%) 7 (8.6%) 6 (6.7%)     Separated/divorced/widowed 18 (32.1%) 18 (22.2%) 18 (20.0%) Age on admission (Mean ± SD) 41.0 ± 11.5 37.0 ± 13.2 38.8 ± 12.1 1.73 (222) 0.18 Axis IV housing problems 34 (60.7%) 35 (43.2%) 36 (40.0%) 6.43 (2) 0.04 Axis IV economic problems 32 (57.1%) 51 (63.0%) 41 (45.6%) 5.40 (2) 0.07 Receiving disability payments over past 3 months (SSI or SSDI) 22 (39.3%) 40 (49.4%) 55 (61.1%) 6.82 (2) 0.03 a 3 2 P F P 2 P Table 3 Clinical and diagnostic characteristics of 227 hospitalized patients Variable a Test statistic (df) P n n n Prior psychiatric hospitalization 36 (65.5%) 58 (71.6%) 75 (84.3%) 7.27 (2) 0.03 Psychotic symptoms on admission 43 (76.8%) 71 (87.7%) 80 (89.9%) 5.20 (2) 0.07 Primary diagnosis, schizophrenia or other psychotic disorder 26 (47.3%) 56 (69.1%) 63 (70.0%) 9.03 (2) 0.01 Comorbid substance use disorder diagnosis 26 (46.4%) 34 (42.5%) 40 (44.4%) 0.21 (2) 0.90 Comorbid personality disorder diagnosis 10 (17.9%) 13 (16.1%) 12 (13.3%) 0.58 (2) 0.75 Change of <20 in GAF score (Discharge–Admission) 23 (42.6%) 26 (32.9%) 47 (53.4%) 7.14 (2) 0.03 Depressive symptoms at discharge 23 (41.1%) 23 (28.4%) 22 (24.4%) 4.69 (2) 0.10 Anxious symptoms at discharge 23 (41.1%) 43 (51.1%) 28 (31.1%) 8.49 (2) 0.01 Psychotic symptoms at discharge 27 (48.2%) 51 (63.0%) 67 (74.4%) 10.34 (2) <0.01 Documented medical problems requiring medication at discharge 35 (62.5%) 41 (50.6%) 32 (35.6%) 10.52 (2) <0.01 >2 Psychiatric medications at discharge 20 (36.4%) 13 (16.1%) 14 (15.9%) 10.40 (2) <0.01 Experiencing side effects at discharge 7 (12.5%) 23 (28.4%) 13 (14.4%) 7.41 (2) 0.02 Required seclusion 5 (8.9%) 11 (13.6%) 18 (20.0%) 3.52 (2) 0.17 Required restraints 4 (7.1%) 8 (9.9%) 8 (8.9%) 0.31 (2) 0.86 Required PRN medicine 14 (25.0%) 30 (37.0%) 38 (42.2%) 4.48 (2) 0.11 Patient has an established outpatient clinician 15 (26.8%) 20 (25.0%) 38 (42.2%) 6.78 (2) 0.03 Good treatment plan adherence during week prior to discharge 41 (73.2%) 60 (74.1%) 50 (55.6%) 8.06 (2) 0.02 Treatment team’s opinion of likelihood of follow-up     Poor 15 (26.8%) 16 (20.0%) 35 (38.9%) 15.02 (4) <0.01     Fair 19 (34.0%) 29 (36.3%) 38 (42.2%)     Good 22 (39.3%) 35 (43.8%) 17 (18.9%) Length of stay in days (Mean ± SD) 11.3 ± 7.5 15.5 ± 8.4 8.4 ± 6.0 20.45 (224) <0.0001 Unit, longer-stay inpatient unit 36 (64.3%) 67 (83.8%) 37 (41.1%) 32.84 (2) <0.0001 a 1 Fig. 1 Average lengths of hospital stay (days) in the three groups of patients  2 17 2 P Fig. 2 Percentage of patients adhering with the first scheduled community mental health appointment following hospitalization  4 Table 4 Independent variables significantly associated with admission and discharge legal status among 221 hospitalized patients Variable a B SE aOR 95% CI Psychotic symptoms at discharge INV-VOL −0.02 0.44 0.98 0.42, 2.32 INV-INV 1.49 0.46 4.42 1.80, 10.86 Documented medical problems requiring medication at discharge INV-VOL −0.48 0.40 0.62 0.28, 1.35 INV-INV −1.51 0.43 0.22 0.10, 0.51 Number of psychiatric medications at discharge (>2) INV-VOL −1.80 0.52 0.17 0.06, 0.46 INV-INV −0.88 0.51 0.41 0.15, 1.12 The model controls for psychiatric hospital unit, length of hospital stay, prior psychiatric hospitalization, and receiving disability payments (SSI/SSDI) a B, Coefficient value from the final regression model; SE, standard error; aOR, adjusted odds ratio; CI, confidence interval Patients who were involuntary on admission and discharge (INV-INV) were approximately 4.4 times more likely to have psychotic symptoms at discharge, relative to the VOL-VOL group. Patients in the INV-INV group were also about 4.5 times less likely (or equivalently, 0.22 times as likely) to have documented medical problems requiring medications at discharge, relative to those in the VOL-VOL group. Patients who were involuntary on admission and voluntary on discharge (INV-VOL) were approximately 6 times less likely (0.17 times as likely) to be prescribed more than two psychiatric medications at discharge. Discussion Multivariate logistic regression modeling yielded three clinical variables that were independently significantly associated with admission/discharge legal status, even after controlling for psychiatric unit, length of hospital stay, prior psychiatric hospitalization, and whether or not the patient received disability payments. As described below, the associations between legal status and each of these three significant variables potentially may be explained by the lower overall treatment engagement common among involuntary patients. First, patients in the INV-INV group were much more likely to be experiencing psychotic symptoms at discharge, compared to the VOL-VOL group. The most plausible explanation for this finding relates to a difference in the level of acceptance of treatment between the two groups. Involuntary legal status is generally an indicator of a lack of treatment engagement, resistance to treatment, and/or impaired insight. Bivariate test results for “treatment plan adherence during the week prior to discharge” support the idea that involuntary patients generally had poorer overall treatment adherence. 3 A significant difference in psychotic symptoms at discharge was not found between the INV-VOL group and the VOL-VOL group. This is not surprising given the fact that these two groups were virtually equally as likely to be rated as having good treatment adherence during the week prior to discharge. Furthermore, these patients were discharged after sufficient resolution of symptoms rather than legal expiration. From a clinical vantage point, it appears that INV-INV patients are being discharged “too soon” due to legal constraints. An alternate explanation for the difference in psychotic symptoms at discharge among the three groups is that INV-INV patients were more likely to have psychotic symptoms at the time of admission compared to the other two groups. However, the bivariate test revealed no statistically significant difference between the three groups in terms of psychotic symptoms present on admission. Second, patients admitted and discharged involuntarily (INV-INV) were much less likely to have documented medical problems requiring medications at the time of discharge, relative to the VOL-VOL group. Because length of stay was controlled for in the model, the most likely reason for the difference in medical problems at discharge is a generalized lack of treatment engagement among involuntary patients. That is, poor treatment engagement is associated with a decreased likelihood of medical problems being reported and diagnosed. Involuntary patients are more likely to refuse medical history taking, physical examination, vital sign checks, diagnostic blood work, and other diagnostic procedures during hospitalization. Thus, clinicians may not have sufficient opportunities or information to detect common medical problems, such as hypertension, diabetes mellitus, and anemia. It is notable that a significant difference in documented medical problems requiring medication was not found between the INV-VOL and VOL-VOL groups. As noted earlier, these two groups had similar treatment adherence during the week prior to discharge. These results support the idea that converting to voluntary status is associated with better treatment engagement, affording more opportunities for medical problems to be detected and medication to be prescribed. Future research is needed to determine whether a general lack of treatment engagement alone, or some additional factors in combination with poor treatment engagement, accounts for the difference in documented medical problems requiring medication. Some alternate explanations as to why the INV-INV patients had fewer medical problems requiring medication at discharge merit consideration. For example, it is conceivable that even when clinicians detect a medical problem, they may be biased toward not prescribing medication to involuntary patients who are highly unlikely to agree to take it. However, in this particular study setting, when medical problems are detected in involuntary patients, clinicians typically prescribe the indicated medication, regardless of the likelihood that the patient will comply with taking it. Another possible explanation is that patients in the INV-INV group are generally healthier than the other two groups. However, because the three legal status groups were very similar with respect to age, gender, and race, it is highly unlikely that one group of patients would have been healthier than the other two. Third, though the INV-VOL and VOL-VOL groups did not differ with respect to psychotic symptoms and medical problems requiring medication at discharge, these two groups did differ in the number of psychiatric medications at time of discharge. Those patients in the INV-VOL group were much less likely to have been prescribed more than two psychiatric medications at discharge, relative to the VOL-VOL group. The difference in number of psychiatric medications between the INV-INV and VOL-VOL groups was not significant, though the effect for the INV-INV group was in the same direction as with the INV-VOL group. The most plausible explanation for this finding may be related to both a lower overall treatment engagement among involuntary patients and differences in clinician prescribing behavior for involuntarily admitted patients compared to voluntary patients. That is, perhaps clinicians err on the side of caution with patients recently converting to voluntary status by limiting the number of psychiatric medications used to treat symptoms. Clinicians may limit the number of prescribed medications, especially if psychiatric symptoms are resolving, in order to simplify the medication regimen and sustain good treatment adherence. Interestingly, a much higher percentage of patients in the INV-VOL group adhered with the first community mental health appointment compared to the VOL-VOL group and the INV-INV group. It is conceivable that a simpler treatment regimen (one or two psychiatric medications) may contribute to higher rates of adherence with initial community mental health follow-up. 11 12 19 20 11 13 11 12 11 5 8 11 21 A second methodological limitation is that commitment statutes “censor” the number of patients that convert to voluntary status, since non-converters must be discharged upon expiration of the commitment. Therefore, the true difference between the INV-VOL and INV-INV groups is somewhat obscured because of the limited timeframe in which patients have to convert to voluntary status. One cannot assume that any given INV-INV patient would never have converted to voluntary status; rather one can only infer that he or she had not done so by time of discharge. In other words, caution must be exercised in assigning ontological status to the INV-INV group. A third limitation is that patients re-admitted during the data collection period were excluded from the analysis. Future research might consider focusing on INV-INV patients with multiple hospitalizations. Another limitation with regard to data collection was that clinical and diagnostic variables (some of which were subjectively rated) were based on clinicians’ reports or information in the hospital chart, without using rating scales administered directly to patients. However, given the goals of this study, the researchers deemed patient assessments, beyond that done as part of routine evaluation and treatment, unnecessary. Also, involuntary patients would have been less likely to participate in formal research evaluations compared to voluntary patients, thus restricting the ability of the analyses to examine the issues of interest. Lastly, the findings from this study may have limited generalizability, given the specific sociodemographic and diagnostic characteristics of the sample. However, the population of interest to the researchers was that of minority individuals with severe psychiatric illnesses being treated in a large, urban, public-sector hospital. Despite the methodological limitations inherent in the study design, there were also several strengths in the methodology and data analysis process. First, an extremely thorough assessment of confounding was conducted. The final model controlled for four variables that were considered likely confounders. Second, alternate modeling techniques were conducted, all of which resulted in the same findings. Lastly, a concerted effort was made to minimize missing values during data collection, and to prevent biases, the few variables with significant missing values were excluded from the analysis. Conclusions 3 8 22 23