Back to Home Page Back to Resume

Factors Affecting Emergency Service Utilization at a Rural Community Mental Health Center

Steven Neil Sobel, M.D.
Steven Anisman, M.A.
H. I. Hamdy, Ph.D.

Originally published in Community Mental Health Journal,
Vol 34, No. 2, April 1998
Pages 157-163

ABSTRACT: Various factors may contribute to cyclical fluctuations in mental health emergency service utilization. Information regarding service utilization patterns would assist the planning and administration of such programs. Five years of data on emergency service contacts at a rural community mental health center were analyzed retrospectively in order to clarify which of an array of potential factors actually affect the utilization of emergency services. One way analyses of variance (ANOVA) or analogous nonparametric tests were performed. Furthermore analyses of significant group differences were performed by means of multiple comparison tests and confidence intervals. Significant associations with number of emergency contacts were found for season, day of week, holidays, school vacations and weather conditions. This study helps to elucidate those factors affecting psychiatric emergency service utilization at a rural community mental health center. Such factors may differ from those affecting other indicators of psychopathology.


Research regarding mental health services utilization is currently of significant interest given the ongoing efforts to control the costs of health care. Various factors may contribute to cyclical fluctuations in the utilization of mental health emergency services. Knowledge about these factors might enable hospital and mental health administrators to better plan and implement effective emergency services. A search for relevant professional literature reveals that there is a surprising paucity of studies examining predictors of emergency service needs throughout the year. Currently, we are left to rely mainly on speculation, if not superstition. Previous studies have generally focused on extreme outcomes, especially suicide. Studies focusing on suicide do not necessarily directly relate to the issue of emergency service utilization which is addressed in this study. By analyzing five years of data on emergency visits, this retrospective study attempts to clarify which of an array of factors actually correlate with mental health emergency service utilization.


This study reviewed psychiatric emergency contacts during a five year period (1988 through 1992) in two rural counties of northwestern Vermont. Emergency services for these counties are provided by the staff of Franklin-Grand Isle Mental Health Service. The contacts may be face-to-face or via telephone and may occur at the clinic, the local general hospital or in the community. Information regarding emergency contacts was retrieved from a computerized database of clinician’s contact forms. Information recorded on these forms includes date, time of day, age and sex. We investigated fluctuations in the mean daily number of adult (age 18 and over) and child emergency contacts during (9:00 a.m.—5:00 p.m.) and after hours with the following factors: day of the week, season school vacations, temperature, occurrence of rainfall or snowfall. We also examined the effect of major holidays (Thanksgiving, Christmas, New Year’s Day) and the weeks preceding and following these holidays on emergency contacts and presence or absence of a full moon on after hours emergency contacts. School vacation dates were provided by a local high school. Lunar cycle information was obtained from the Fairbanks Museum and Planetarium in St. Johnsbury, VT. Data regarding daily high and low temperatures and amount of rainfall or snowfall were obtained from a local radio station (WWSR). We performed a series of tests to determine whether or not each factor examined was associated with significant differences (p <0.05) in total number of contacts, adult contacts, child contacts, during hours contacts and after hours contacts. Either the usual one way analyses of variance (ANOVA) or analogous nonparametric tests (e.g. Kruskal-Wallis, Mann-Whitney and two-tailed Wilcoxon signed rank test) were employed, depending upon the validity of the normality assumption. (The Kolmogrov-Smirnov Goodness of Fit test was used to justi~r this assumption). Significant results were further analyzed to identify the significant group differences by means of multiple comparison tests (e.g. Duncan,Tukey-HSD,and LSD tests) and the use of confidence intervals. The median test and chi-square analyses were also utilized when appropriate. The Pearson correlation coefficient was utilized to determine the strength of linear relationship between number of contacts during hours with the number of contacts after hours.


There were a total of 13,974 emergency contacts over the five year period. Fifty eight percent of patients were female and 42% were male (for 389 contacts this information was not recorded).The average age was 34.5 years. Adult contacts numbered 10,770 (77%) and child contacts numbered 3204 (23%). The local population of northwestern Vermont is a markedly homogeneous Caucasian one and the data on ethnic groups reflected this relative lack of diversity (99% Caucasian).
     Statistical analyses were performed to examine the correlations of the various factors with the mean daily number of total contacts, adult and child contacts and during and after hours contacts. Table 1 shows the p-values for these comparisons.
     T-tests indicated that the overall mean daily number of during hours contacts (4.83) was significantly different (d.f. 3358, p = 0-000) from the mean number of after hours contacts (2.82).
     The total number of contacts was highest in winter and lowest in summer. The decrease in total summertime contacts is apparently explained by the highly significant (p = 0.000) drop in the number of children seen during that season. Child contacts were highest in the fall. With regard to weekdays, the number of contacts during hours was greater on Mondays and Tuesdays (one-way ANOVA with p = 0.000). On the other hand, there were more after hours contacts on Fridays (according to LSD and Duncan significance tests but not according to Tukey-HSD test). Total contacts were markedly decreased on weekends.
     On the day of a major holiday, significantly fewer total and adult emergency contacts occurred. This was also found to be the case for the week preceding and the week following the major holiday. Additionally, child contacts decreased during the holiday weeks.
     Lunar phase (full moon or not) had no association with the number of emergency contacts (even when examined only for after hours contacts).
     With regard to weather conditions, the occurrence of rainfall was associated with increased number of emergency contacts among children only. Snowfall predicted significantly more total and during hours contacts, but the level of significance of increased contacts was greatest among children (p 0.000). Higher temperatures (> 81 degrees Fahrenheit) were associated with fewer total, child and daring hours contacts. Wintertime temperatures below zero degrees Fahrenheit were associated with fewer total contacts and fewer during hours contacts.
     The total number of calls was less during school vacations as was the number of child contacts specifically. Interestingly, even the number of adult contacts decreased during school vacations (p = 0.01). This decrease in the number of contacts was not significant when only after hours contacts were considered.
     The number of contacts after hours was (very weakly) negatively. correlated with the number of contacts during hours (Pearson correlation coefficient of —0.16 at a significance level of p = 0.000).


Effect of Different Factors on Number of Contacts
FactorsTotal Number of ContactsDuring Hours ContactsAfter Hours ContactsChild ContactsAdult Contacts
Full Moon0.4160.8980.1500.9680.343
Holidays week0.000*0.001*0.0700.016*0.000*
School vacation0.000*0.000*0.1790.000*0.011*
Low temp.0.002*0.046*0.2240.000*0.104
High temp.0.012*0.026*0.007*0.000*0.592
* means the factor is significant at x=5% and N/A means not applicable.


This study helps to elucidate those factors predicting psychiatric emergency service utilization at a rural community mental health center. Few previous studies have addressed this specific issue. Such factors may differ from those affecting rates of suicide, depression or other indicators of psychopathology.
     Others have found a seasonal effect on suicide rates (Lester, 1979; Eastwood & Peacocke, 1976; Souetre, Salvati, Belugou, Douillet, Braccini & Darcourt, 1987; Wenz, 1977; Lester, 1971; McCleary, Chew, Hellsten & Flynn-Bransford, 1991; Maes, Cosyns, Meltzer, De Meyer & Peeters, 1993) although the particular Season identified as being associated with increased suicide rates has not been consistent across studies. Our finding of increased total number of contacts in the winter is consistent with evidence regarding seasonal changes in rates of depression (Wehr & Rosenthal, 1989; Maes, Meltzer~ Suy, & De Meyer, 1993). The most prominent seasonal effect which we found was fewer child emergency contacts in the summer. This is probably due to the fact that school referrals account for many of the child emergency contacts during the rest of the year.
     The increase in emergency contacts during hours on Mondays and Tuesdays could be due to the fact that people who call in or are seen on the weekend are often scheduled to be seen by the emergency service worker following the weekend. Such follow-up sessions may not represent true emergencies. The lack of such scheduled “emergency” visits on the weekend most likely explains the dramatic decrease in total, adult and child contacts on Saturdays and Sundays. On the other hand, the increase in emergency after hours contacts on Fridays might suggest that for many individuals in treatment at a community mental health center, the prospect of less intense follow-up over the weekend may lead to increased utilization of emergency services.
     The finding that emergency contacts decreased during holidays and holiday weeks contradicts popular notions, but is in agreement with much professional literature on decreased suicide rates (Lester, 1979; Sonetre et al., 1987; Zung & Green, 1974; Jones & Jones, 1977) and decreased psychopathology (Hillard, Holland & Ramin, 1981) during the holiday season. However, our findings are probably again partly explained by a decrease in scheduled visits with emergency staff during these times, either due to the mental health center being closed on the day of a holiday for nonemergency visits or due to more staff members being on vacation at these times (holiday weeks). Thus, to some extent this could be an artifact of a system in which not all visits to the emergency service represent true emergencies.
     A popular belief persists that a full moon means a busy night with regard to psychiatric emergencies. However, as expected, this was shown to have no basis in fact, which is consistent with the literature other regarding suicide rates (Lester, 1979; Maldonado & Kraus, 1991).
     Interestingly, the number of contacts with children appeared to vary most dramatically with different weather conditions (precipitation and temperature). It is not clear to us why this would be so or why contacts would increase with rain or snow. Perhaps, when weather conditions keep children indoors, parents may be more directly confronted with problem behaviors. Extremely cold weather does appear to deter people from venturing out even to address emergency psychiatric needs. High temperatures in summertime were not associated with an increase in emergency contacts. However, we should note that extreme high temperatures are unusual in Vermont. Thus the findings might be different in geographical areas with a significantly higher upper range of temperatures.
     It is not surprising that child emergency contacts decreased during school vacations as many such contacts are initiated by school referrals. This explanation is supported by the finding that only the number of contacts during working hours was affected. It is interesting that adult contacts also showed a decrease. This could be explained by the fact that adults who are parents might find it more difficult to leave home (e.g. due to lack of child care arrangements) during school vacations. The correlation coefficient of —0.16 indicates a lack of a relationship between the number of emergency contacts during hours with the number after hours.
     Conclusions regarding emergency service utilization cannot be based upon extrapolations from research on other parameters of psychopathology such as suicide. Research which specifically and directly explores this issue is needed. The large number of emergency contacts and the relatively long follow-up period are strengths of this study. However, the retrospective nature of the data and lack of detailed information regarding the clinical circumstances of each individual emergency contact are limitations. Future research employing a prospective design could further contribute to our understanding of the factors affecting emergency service utilization.


     Steven Neil Sobel, M.D., is a Staff Psychiatrist and Steven Anisman, M.A., is an Emergency Service Clinician at the Franklin-Grand Isle Mental Health Service. H. I. Hamdy, Ph.D., is affiliated with the Department of Biostatistics, University of Vermont, Burlington, VT.
     Address correspondence to Steven N. Sobel, M.D., Franklin-Grand Isle Mental Health Service, 8 Ferris Street, St. Albans, VT 05478.
     The authors wish to thank Edward Giroux and William Post for their technical assistance in accessing the computer data base.

Back to Top of Page Back to Resume Back to Home Page

Last Updated: