Hospitalization in the First Year of Treatment for Schizophrenia

Objective: To determine the rates of hospitalization during the first year of treatment for schizophrenia, using an epidemiologic sample.

Method: We examined inpatient and outpatient administrative databases in the province of Nova Scotia for cases of schizophrenia (ICD-9 code 295 or 298) newly diagnosed during the years 1995 to 1998. We noted the diagnosis site (that is, inpatient or outpatient) and hospitalizations in the year following diagnosis. We also established links to the clinical database maintained by the Nova Scotia Early Psychosis Program (NSEPP).

Results: Over the 4-year period, we identified 434 unique cases from an at-risk population of 320 000 (yielding a yearly average age-specific incidence rate of 3.3/10 000), of whom 119 had received care from the NSEPP. Of the cases, 54% were initially diagnosed while they were inpatients. In the year following diagnosis, the overall hospitalization rate, excluding initial hospitalizations, was 17%. Patients who were initially diagnosed while inpatients had a higher rate of hospitalization in the first year of treatment (25% vs 7%), compared with those initially diagnosed while outpatients. This relation was also present among patients who received care from the NSEPP.

Conclusions: Of newly diagnosed patients with schizophrenia, 46% were not hospitalized at the time of initial diagnosis. Of all patients, 17% required hospitalization during the first year of treatment, excluding an initial hospitalization, if present. Hospitalization rates in the first year were higher among patients initially hospitalized and among those with a rural residence. Patients requiring hospitalization during the first year form an important target group for improved interventions.

(Can J Psychiatry 2004;49:635-638)

Information on funding and support and author affiliations appears at the end of the article.

Clinical Implications

* Approximately one-half of newly diagnosed patients with schizophrenia do not require inpatient services.

* Patients who are first diagnosed while inpatients and those who reside in rural areas are more likely to require additional inpatient services in the first year of treatment.

* Patients who require hospitalization in the first year of treatment may require targeted interventions to improve outcomes.

Limitations

* Data were drawn from databases established for administrative rather than research purposes.

* Some patients who received previous treatment outside Nova Scotia may have been included.

* Several factors, including illness severity and availability of appropriate outpatient services, are likely to influence hospitalization rates.

Key Words: schizophrenia, early psychosis, hospitalization rates

Over the past decade, interest has increased in designing services to optimize care during the early stages of psychotic disorders (1,2). In designing these services, it will be important to study such service indicators as hospitalization rates and to investigate factors that may influence these indicators.

The province of Nova Scotia is an advantageous setting in which to carry out such studies. Available administrative databases contain all inpatient admissions and outpatient psychiatric contacts. As do all Canadian provinces and territories, Nova Scotia provides universal access to health care.

A prototype Early Psychosis Program, designed to enhance prompt assessment and optimize treatment, has been operating in the province since late 1995 (3). The program has focused on education about the signs and symptoms of early psychosis (4) while also providing expert clinical care for a self-selected convenience sample of first-episode patients (5).

This study measured hospitalization rates during the first year of treatment for schizophrenia in Nova Scotia and identified factors that might influence those rates.

Methods

We accessed administrative health data for the province of Nova Scotia (population 940 000) through the Population Health Research Unit at Dalhousie University, Halifax. At the time of analysis, physician and hospital administrative databases covering the period January 1, 1990, to December 31, 1999, were available. Encrypted health card numbers allowed data linkages while ensuring confidentiality. We retained data for Nova Scotia residents aged 15 to 3 5 years (an at-risk population of 320 000) with a diagnosis of schizophrenia (ICD-9 codes 295 or 298) made between 1995 and 1998, either in hospital or by an outpatient psychiatrist, and with no recorded diagnosis of schizophrenia in the 5 years preceding the year of diagnosis. Physician and hospital administrative databases were cross-checked to determine whether the first diagnosis was made in an inpatient or outpatient setting. Location of residence allowed cases to be identified as residing either in the only urban centre, Halifax (population 360 000), or elsewhere. Linkage with databases maintained by the Nova Scotia Early Psychosis Program (NSEPP) allowed us to identify patients who received care from that program. We identified hospitalizations in the first year after diagnosis where schizophrenia was the primary diagnosis.

We used logistic regression to determine the odds of having at least 1 hospitalization in the year following diagnosis when we controlled for age, sex, location of residence, and initial diagnosis site (that is, an inpatient vs an outpatient setting). We obtained research ethics approval for the study from the Ethics Review Board of the Capital District Health Authority, Halifax, Nova Scotia.

Results

We identified 434 new cases over the 4-year study period, of whom 95% were assigned ICD-9 code 295. Given an at-risk population of approximately 320 000, the yearly average incidence rate was 3.3/10 000.

Table 1 summarizes demographic characteristics and hospitalization rates. Of the patients, 236 (54%) were hospitalized at the time of initial diagnosis. Excluding hospitalization at the time of diagnosis, 74 patients (17%) were hospitalized at least once in the year after their initial diagnosis.

Table 1 also compares patients first diagnosed while they were inpatients with those first diagnosed as outpatients. The groups did not differ significantly with regard to age. However, compared with outpatients, inpatients were significantly more likely to be women, were more likely to reside outside the urban area, and were less likely to have had contact with the NSEPP. Finally, inpatients were more likely to be hospitalized in the year following diagnosis (25% vs 7% for outpatients).

Using the variables in Table 1, results from a logistic regression model for the probability of having at least 1 hospitalization in the first year after diagnosis demonstrated that sex and age at diagnosis were not significantly associated with the probability of hospitalization. Residence outside the urban area (odds ratio 1.71) and initial diagnosis while an inpatient (odds ratio 2.87) were significantly associated with a higher probability of hospitalization in the first year.

Of the 434 patients, 119 received care from the NSEPP. Compared with the entire sample, these patients were significantly younger, were more likely to have received their first diagnosis as outpatients, and were less likely to be hospitalized in the year following diagnosis. The hospitalization rate in the first year after diagnosis was 8%. In a logistic regression model that also included the site of initial diagnosis, having received care from the NSEPP did not significantly determine hospitalization in the first year after diagnosis.

Discussion

Three findings emerge from this study. First, nearly one-half of the patients were not hospitalized during the process of initial diagnosis. second, excluding an initial hospitalization, slightly less than one-fifth of the patients required hospitalization during the first year of treatment after diagnosis. Third, hospitalization in the first year of treatment was higher for patients first diagnosed in hospital and for patients who lived outside the single urban area in Nova Scotia.

Before discussing these points, 2 methodological issues should be noted. First, the databases used were established for administrative rather than research purposes. As has been recently noted (6), administrative databases provide a cost-effective means for studying the health of specific populations. Conversely, they have limitations. For example, although all the diagnoses were made by a psychiatrist, diagnostic interrater reliability cannot be established. As well, it should be noted that the databases use ICD-9 coding, rather than the DSM-IV criteria with which Canadian clinicians are more familiar. A second limitation involves the possibility that some patients who were first diagnosed in another jurisdiction and later moved to Nova Scotia could have been mistakenly identified as new cases. That said, the incidence rate determined is very similar to that reported elsewhere (7), suggesting there were few such cases.

We derived 2 different hospitalization rates from the current results. The first rate is the percentage of patients hospitalized at the time of first diagnosis. The second rate is the percentage of patients hospitalized during the first year following diagnosis, excluding the initial hospitalization, if there was one.

As to the first rate, the current results indicate that 54% of the patients were hospitalized at the time of first diagnosis. Very similar rates have been reported from the UK (8), from Jamaica (9), and from London, Ontario (10). These results emphasize the point that nearly one-half of all patients with schizophrenia are diagnosed and begin treatment without requiring an inpatient admission.

The second rate measured in this study-hospitalization during the first year of treatment-provides a measure of treatment response. The results yield an overall rate of 17%; that is, 17% of all patients were hospitalized at some point during the first year of treatment, excluding the initial hospitalization, if there was one. Again, very similar rates have been reported from Jamaica (9) and from London, Ontario (10). These figures indicate that, with current treatment, a significant number of patients will need access to inpatient service during their first year of treatment.

The finding that the hospitalization rate during the first year of treatment was higher for patients hospitalized at the time of their initial diagnosis and for those who resided in a rural location requires discussion and further research that goes beyond the scope of this study. Several factors, including severity of presentation, diagnostic uncertainty, and comorbid disorders, are likely to influence hospitalization rates (11).

Given the limitations of the administrative databases used in this study, it is difficult to determine the reasons for the observed differences. Regression analysis indicated that both the initial diagnosis site and urban vs rural residence were independently associated with differences in hospitalization rates. Because the differences in hospitalization rates were present among the subset of patients seen by the NSEPP, differences in clinical practice alone are unlikely to explain this finding.

One likely explanation is that patients hospitalized at the time of diagnosis differ clinically from those diagnosed as outpatients, particularly in terms of the severity of their presentation. This may indicate that the hospitalized patients have a more severe form of disorder and, hence, do not respond as well to the available treatment. Unfortunately, the administrative databases we used in this study do not provide information on clinical severity, and we could not test this explanation. Likewise, the databases do not contain accurate data on comorbid psychiatric or physical disorders that might contribute both to the need for hospitalization and to poorer treatment response.

Regardless of explanations, our study results focus attention on the group of patients who require hospitalization in their first year of treatment. If we assume that the need for hospitalization was associated with a relatively poor treatment response, these patients-17% of the total sample and 25% of initially hospitalized patients-represent a significant subset for whom current treatments are not sufficiently effective. Targeted interventions for these patients could be a useful focus for further research and service development.

Funding and Support

Funding for this study was provided in part by a developmental grant from the Nova Scotia Health Research Foundation.

Acknowledgements

We gratefully acknowledge the support of our collaborators in the Population Health Research Unit at Dalhousie University, in particular, Mike Pennock.

R�sum� : L'hospitalisation dans la premi�re ann�e de traitement de la schizophr�nie

Objectif : D�terminer les taux d'hospitalisation durant la premi�re ann�e de traitement de la schizophr�nie, � l'aide d'un �chantillon �pid�miologique.

M�thode : Nous avons examin� les bases de donn�es administratives des patients hospitalis�s et externes de la province de la Nouvelle-Ecosse pour trouver les cas de schizophr�nie (CIM-9 code 295 ou 298) nouvellement diagnostiqu�s de 1995 � 1998. Nous avons not� l'endroit du diagnostic (c.-�-d., patient hospitalis� ou externe) et les hospitalisations dans l'ann�e suivant le diagnostic. Nous avons aussi �tabli des liens aux bases de donn�es cliniques entretenues par le programme de psychose pr�coce de la Nouvelle-Ecosse (NSEPP).

R�sultats : Sur une p�riode de 4 ans, nous avons trouv� 434 cas uniques sur une population � risque de 320 000 (ce qui donne un taux d'incidence moyen annuel selon l'�ge de 3,3/10 000), sur lesquels 119 avaient re�u des soins du NSEPP. Parmi les cas, 53 % ont �t� diagnostiqu�s la premi�re fois tandis qu'ils �taient hospitalis�s. Dans l'ann�e suivant le diagnostic, le taux d'hospitalisation global, en excluant la premi�re hospitalisation, �tait de 17 %. Les patients qui ont �t� diagnostiqu�s la premi�re fois tandis qu'ils �taient hospitalis�s avaient un taux d'hospitalisation plus �lev� dans la premi�re ann�e de traitement (25 % c. 7 %), comparativement � ceux diagnostiqu�s la premi�re fois en tant que patients externes. Cette relation �tait aussi pr�sente chez les patients qui recevaient des soins du NSEPP.

Conclusions : Sur les patients nouvellement diagnostiqu�s schizophr�nes, 46 % n'�taient pas hospitalis�s au moment du diagnostic initial. De tous les patients, 17 % ont eu besoin d'une hospitalisation durant la premi�re ann�e de traitement, en excluant la premi�re hospitalisation, le cas �ch�ant. Les taux d'hospitalisation de la premi�re ann�e �taient plus �lev�s chez les patients hospitalis�s lors du diagnostic initial et chez ceux demeurant en milieu rural. Les patients ayant eu besoin d'une hospitalisation durant la premi�re ann�e de traitement forment un important groupe cible pour des interventions am�lior�es.

[Reference]

References

1. McGorry PD, Jackson HJ. The recognition and management of early psychosis: a preventative approach. Cambridge (UK): Cambridge University Press; 1999.

2. Birchwood M, Fowler D, Jackson C. Early intervention in psychosis: a guide to concepts, evidence and interventions. Chichester (UK): Wiley & Sons; 2000.

3. Whitehorn D, Lazier L, Kopala L. Psychosocial rehabilitation early after the onset of psychosis. Psychiatr Serv 1998;49:1135-7.

4.Crown MJ, Milliken H, Whitehorn D, Alexiadis M, Kopala L. Developing an education resource package to facilitate early detection and optimal treatment: The Sooner the Better Project. Schizophr Res 2003;60:333.

5. Whitehorn D, Brown J, Richard J, Rui Q, Kopala L. Multiple dimensions of recovery in early psychosis, Int J Psychiatry 2002;14:273-83.

6. Goldner EM, Jones W, Waraich P. Using administrative data to analyze the prevalence and distribution of schizophrenia disorders. Psychiatr Serv 2003 ;54:1017-21.

7. Bresnahan MA, Brown AS, Schaefer CA, Begg MD, Wyatt RJ, Susser ES. Incidence and cumulative risk of treated schizophrenia in the prenatal determinants of schizophrenia study. Schizophr Bull 2000:26:297-308.

8. Sipos A, Harrison G, Gunnell D, Amin S, Singh SP. Patterns and predictions of hospitalization in first-episode psychosis. Br J Psychiatry 2001 ; 178:518 -23.

9. Hickling FW, McCallum M, Nooks E, Rogers-Johnson P. Outcome of first contact schizophrenia in Jamaica. West Indian Med J 2001;50:194-7.

10. Malla AK, Norman RM, Manchanada R, McEean TS, Harricharan R, Cortese L, and others. Status of patients with first-episode psychosis after one year of phase-specific community-oriented treatment. Psychiatr Serv 2002;53:458-63.

11. Sridharan B, Arsah P, Marcos M. Hospitalisation in first-episode psychosis. Br J Psychiatry 2002;180:84-5.

[Author Affiliation]

David Whitehorn, PhD, MScN1, Julie C Richard, BSc2, Lili C Kopala, MD, FRCPC3

[Author Affiliation]

Manuscript received July 2003, revised, and accepted October 2003. Previously presented in part at the 8th International Congress on Schizophrenia Research; 2001 April 28-May 2; Whistler (BC).

1 Clinical Nurse Specialist, Nova Scotia Early Psychosis Program, Capital District Mental Health Program, Dartmouth, Nova Scotia; Lecturer, Department of Psychiatry, Dalhousie University, Halifax, Nova Scotia.

2 Formerly, Research Assistant, Nova Scotia Early Psychosis Program, Capital District Mental Health Program, Dartmouth, Nova Scotia; Currently, Medical student, University of Ottawa, Ottawa, Ontario.

3 Formerly, Founding Director, Nova Scotia Early Psychosis Program, Capital District Mental Health Program, Dartmouth, Nova Scotia; Currently, Clinical Professor, Center for Complex Brain Disorders, Department of Psychiatry, University of British Columbia, Vancouver, British Columbia.

Address for corresponce: Dr D Whitehorn, Nova Scotia Early Psychosis Program, 300 Pleasant Street, Dartmouth, NS B2Y 3Z9

e-mail: david.whitehorn@cdha.nshealth.ca