This article was downloaded by: [Indiana University Libraries] On: 16 March 2015, At: 22:30 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of American College Health Association Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/vzch20
Factors Affecting the Usage Rate for Mental Health Sections of College and University Health Services Allan J. Schwartz Ph.D. a
a b
University of Rochester, School of Medicine and Dentistry , USA
b
University Health Service, University of Rochester , 250 Crittenden Blvd., Box 617, Rochester, New York, 14642, USA Published online: 07 Apr 2011.
To cite this article: Allan J. Schwartz Ph.D. (1979) Factors Affecting the Usage Rate for Mental Health Sections of College and University Health Services, Journal of American College Health Association, 28:3, 140-144, DOI: 10.1080/01644300.1979.10392917 To link to this article: http://dx.doi.org/10.1080/01644300.1979.10392917
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Factors Affecting the Usage Rate for Mental Health Sections of College and University Health Services ALLAN J . SCHWARTZ, Ph.D.* University o f Rochester
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
Abstract The history o f the Mental Health Annual Program Survey (MHAPS) i s briefly reviewed. Characteristics o f the institutions participating in MHAPS during the 1975-76 program year are reported. Usage rate data for the sample are reviewed, and the procedures employed in determining the predictors o f sample variability in mental health section usage rate are described. Variability in usage rate was regressed against six categories o f variables: institutional characteristics, alternative available resources, mental health section staffing levels and staffing patterns, allocations o f staff time, and programming options. When the significant predictors included in these six categories were pooled, five variables were sufficient t o account for 95% o f usage rate variance. The number o f weekly interview hours per 1000 full-time students and the proportion o f full-time staff each accounted for 29% o f usage rate variance. Visits per patient accounted for 18%, time allocated t o consultative services for lo%, and the availability o f off-campus resources for 8%. The effects that changes in the values o f the predictors can be expected t o have on usage rate are illustrated, and the implications o f these results for the use o f comparative data are discussed. Introduction In January 1972, a t a conference in Boulder, Colorado sponsored by the Western Interstate Commission on Higher Education, a decision was made to establish a systematic, annual survey o f the operations o f the mental health sections o f college and university health services.’ Originally named the Student Mental Health Data Bank, the program was renamed the Mental Health Annual Program Survey (MHAPS) in 1973. Since i t s inception, MHAPS has been conducted under the sponsorship o f the American College Health Association. The substantive areas covered by the survey include amount and modes o f service, staffing levels and patterns, complementary formal mental health programs, suicide, official leaves o f absence due t o illness, and allocation o f professional time. Selected data from the first survey, covering the 1970-71 operating year, have been published. Unpublished summaries reviewing data for the 1973-74 and 1975-76 operating years have been distributed t o survey respondents through the American College Health A~sociation.3~4 This report focuses on usage rate data from the 1975-76 survey.
Modally, the sample consisted o f publicly funded (61%), doctorate-awarding (74%) universities (83%) with a median full-time enrollment o f 14,000. The range o f enrollment was quite large, however, with enrollments as small as 1700 and as large as 45,000 being represented. The median proportion o f students living in universityowned or operated housing was 37% o f full-time student enrollment. Sixty-one percent o f the institutions had two or more on-campus agencies that complemented the mental health section, while only 13% had no other on-campus resource for students. The most usual complementary agency was a counseling center. Off-campus mental health resources were most often designated as ample (44%) or moderate (35%), and 39% reported having a medical school on-campus. The geographic distribution o f the sample showed 61% o f the institutions located in New England and in other states bordering the Atlantic. Only one institution was located in the Rocky Mountain and Pacific Coast states, and 35% were located in states lying between the Appalachian and Rocky Mountains.
The Sample o f Respondents
Usage Rate
Twenty-eight institutions responded t o requests for survey data for the 1975-76 operating year, with five o f those reporting that they did not have a designated mental health staff. Based upon past surveys, it i s estimated that about 100 o f the 400 institutions solicited had a designated mental health staff. Accordingly, the sample o f 23 respondents upon which this report i s based represents a return rate o f 20 to 25%.
Usage rate, defined as the number o f different patients seen for each 1000 full-time students enrolled per 12 month operating year (pts/l OOOfts), was derived from patient and enrollment data provided by the reporting institutions. The enrollment figures supplied were for the fall term (semester or quarter) o f the 1975-76 operating year. The 21 institutions providing both patient and enrollment data represented 309,413 full-time students and provided services to 13,753 different patients. For the sample as a whole, therefore, the usage rate was 44pts/l000fts. There was notable variability in usage rate across institutions. The mean o f the 21 individual usage rates, 54pts/lOOOfts, was
*Assistant Professor of Psychiatry (Psychology), University of Rochester School of Medicine and Dentistry, and Staff Psychologist, University Health Service, University of Rochester, 250 Crittenden Blvd., B o x 617, Rochester, New Y o r k 14642
140
J.A.C.H.A.
USAGE R A T E
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
somewhat higher than the median o f 40pts/l OOOfts, largely as a function o f relatively high usage rates a t two institutions with relatively small enrollments (1 7OOfts and 6900fts). Usage rates for these institutions were 179 and 115 respectively, values well above the range within which the remainder o f the sample fell (10-90pts/lOOOfts). The mean, median and range o f usage rates for this sample are quite consistent with those reported in previous studies o f usage rates for college mental health sections.5 The wide variability in usage rate among this sample o f institutions stimulated a search for the correlates and possible determinants o f usage rate. Six categories of variables that seemed likely t o influence usage rates were selected. These categories and the variables initially included in each are listed in Table 1. The decision t o include a particular variable in this preliminary phase was based upon a variety o f criteria including that variables’s intrinsic interest as a predictor, i t s distributional characteristics, and the absence of missing data.
Method Using the statistical package for the Social Sciences (SPSS) programming package, each category o f variables was entered into a separate stepwise multiple regression analysis.6 This permitted identification o f the most parsimonious subset o f statistically significant predictors within each category. Additional analyses were then done using only those variables that had been statistically significant predictors in the preceding analyses. For each category, each variable was entered as the last predictor o f the set t o determine the unique usage rate variance accounted for by that variable. Then the common variance, that is, the proportion o f usage rate variance accounted for by either o f two (or more) variables in each s e t o f predictors was assessed.? In a final set o f analyses, predictors from all six categories were entered into the regression analyses t o determine the single best subset of predictors and t o determine the proportions o f unique and o f common variance accounted for by these variables. Results
Table 1 PREDICTOR V A R I A B L E S USED IN T H E REGRESSION A N A L Y S E S C A T E G O R Y 1: MOSAIC OF A V A I L A B L E RESOURCES Availability o f Off-Campus Services Medical School o n Campus Number o f Special Formal Programs Number o f Other On-Campus Agencies Providing Related Services Percent o f T o t a l On-Campus Weekly Interview Hours Represented By Mental Health Section Interview Hours C A T E G O R Y 2: I N S T I T U T I O N A L C H A R A C T E R I S T I C S Size (full-time enrollment) Institutional Designation (college vs. university) Funding (public vs. private) Percentage o f Students Living O n Campus Highest Degree Awarded Rate o f Official Leaves o f Absence f o r Mental Health Reasons C A T E G O R Y 3: A L L O C A T I O N OF PROFESSIONAL E F F O R T ( T I M E ) Percent in Direct Services Percent in Indirect Services Percent in Consultative Services Percent i n Formal Academic Instruction Percent in Providing Clinical Supervision Percent i n Receipt o f In-Service Training Percent in Doing Research C A T E G O R Y 4: S T A F F I N G PATTERNS Percent F u II-Time Percent w i t h Staff Status (vs. trainees) Percent w i t h Psychiatry Background Percent w i t h Psychology Background Percent w i t h Social Work Background Percent w i t h Nursing Background CATEGORY 5: STAFFING L E V E L Number o f Full-Time Equivalent Personnel Per 1000 Full-Time Students Number o f Weekly Interview Hours Per 1000 Full-Time Students C A T E G O R Y 6: P R O G R A M M I N G OPTIONS Percent o f T o t a l Services = Direct Services Percent o f Outpatient Services = Group Services Percent o f Outpatient Services = Couples/Family Services Visits Per Patient Percent o f Patients Referred f o r Further Care
VOL. 28, DECEMBER 1979
Each category o f variables except programming options was found t o account for a statistically significant proportion o f usage rate variance. As shown in Table 2, the category accounting for the greatest proportion o f usage rate variance, 7376, was the set o f variables that described the mosaic of mental health resources available to students. This result suggests that the presence of alternative resources (both on and off-campus), the existence o f specialized formal programs (such as pregnancy counseling, crisis phone services, and courses in human sexuality), and the proportion o f on-campus resources represented by the mental health section all affect usage rate. Mental health section usage rate was lower when the total number o f specialized formal programs was high, and usage rate was higher when there was a medical school on campus, when off-campus resources were more readily available, and when the mental health section represented a higher proportion o f on-campus resources. Fifty-seven percent o f usage rate variance was accounted for by the second ran king cateogry, institutional characteristics. Offical leaves o f absence for mental health reasons was the single best predictor in this category. This variable alone accounted for 34% o f usage rate variance with a higher rate of leaves predictive o f higher usage rate. When the proportion of students living in university-owned or operated housing was high, so was usage rate. The final significant predictor in this category, highest degree awarded, was negatively related t o usage rate. That is, institutions awarding doctorates tended t o have lower usage rates than those awarding master’s degrees. Allocation o f staff effort (time), the third ranking category, accounted for 53% of usage rate variance. The allocation o f a larger proportion o f staff time to indirect and consultative services was associated with lower usage rate. Greater staff involvement with research was associated with a higher usage rate. In accounting for statistically significant portions o f usage rate variance, staffing patterns was the fourth ranked category (32%), with staffing level fifth and last (29%). TWO staffing pattern parameters, the proportion o f staff designated
tSuppressor effects were also noted f o r variables i n five o f the six categories and f o r the pooled s e t o f variables. These effects indicate t h a t the value o f one predictor ( X i ) is affecting a second predictor ( X 2 ) w i t h the result t h a t removing the effect o f X i on X 2 changes the correlation between X 2 and Y. More technically, the zero order correlat i o n [ryx2] is n o t equal t o the semipartial correlation [ r y ( x 2 . x 1 ) ] .
141
COLLEGE HEALTH
Table 2 USAGE RATE VARIANCE ACCOUNTED FOR BY VARIABLES I N EACH O F SIX CATEGORIES Percent Of Total Usage Rate Variance Accounted For1
Direction of Association
1. MOSAIC OF A V A I L A B L E RESOURCES Availability of Off-Campus Services Medical School On Campus (l=yes, 2=no) Number o f Special Formal Programs Percent o f Total OnCampus Weekly Interview Hours Represented by Mental Health Section
73 41 a 7a 9a 35a
positive negative negative Dositive
2. IN ST ITUTION A L CH A RACTE R ISTl CS Percent o f Full-Time Students Eligible for Mental Health Services Rate o f Official Leaves o f Absence for Mental Health Reasons Highest Degree Awarded
57 18a
positive
.05
34
positive
.o 1
10
negative
.01
3. ALLOCATION OF PROFESSIONAL EFFORT (TIME) Percent o f A l l Service Time in Indirect Services Percent o f A l l Service Time in Consultative Services Percent o f Total Time Doing Research
36 6 8 10
negative negative positive
.05 .05 .05
4. STAFFING PATTERNS (base = total full time equivalents of professional personnel) Percent Fu II-T ime Percent with Psychology Background
32 11 14
positive positive
.05 .05
5. STAFFING LEVEL Number of Weekly Interview Hours Per 1000 Full-Time Students
29 29
positive
.05 .05
6. PROGRAMMING OPTIONS Visits Per Out-Patient Percent o f Patients Referred for Further Care
26 14a 13a
negative negative
.10 .10 .10
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
Category and Variables
Significance Level
.o 1 .o 1 .01
.o 1 .01
.o 1
.01
.05
1: Values for categories = square o f multiple correlation. a: Percentage o f variance accounted for reflects supressor effects.
full-time and the proportion with psychology training (vs. psychiatry, social work, nursing) were positively and significantly related to usage rate. In terms o f staffing level, both the number o f full-time equivalent staff per 1000 full-time students (fts) and the number o f weekly interview hours per 1000fts were positively related to usage rate. It i s apparent from summing the proportions o f usage rate variance accounted for by each category o f variables that sizable portions of the variance accounted for by variables in one category would also be accounted for by variables in other categories. This was confirmed when categories were pooled and the s e t o f twelve previously significant variables, augmented by two nearly significant programming option variables, were entered into a stepwise regression analysis. Rate o f mental health leaves, proportion o f personnel with psychology background, proportion o f time spent in research, visits per patient, proportion of all service time devoted t o consultative service, weekly interview hours per 1OOOfts, and proportion o f personnel with full-time commitment were the predictors selected. However, when the unique contribution o f each variable was assessed in relation t o the remaining six, rate o f leaves and time in research were found t o make unique contributions o f less than 1%. This provided both dramatic and specific confirmation o f the redundancy that had been anticipated. Rate o f leaves and time spent in research were removed from the set o f 14 predictors, and a second s e t o f regression analyses was performed. Table 3 summarkes the results o f 142
these analyses. Five variables accounting for 95% o f usage rate variance now entered the regression equation. These were weekly interview hours per 1OOOfts, proportion o f personnel with full-time commitment, number o f visits per outpatient, proportion o f service time devoted to consultative services, and the availability o f off-campus mental health services. Each o f these variables provided a unique and statistically significant increment in usage rate accounted for, within the final set o f five predictors. The stuffing level and stuffing pattern variables (weekly interview hours per 1OOOfts and proportion o f full-time personnel) each accounted for 29% o f usage rate variance. Visits per out-patient accounted for 17%, proportion o f time devoted to consultative services for lo%, and availability o f off-campus resources for 8%. Only 2% o f the composite 95% o f usage rate variance accounted for was common or overlapping variance. Visits per outpatient and proportion o f service time devoted to consultative services were negatively associated with usage rate, while the other three variables were positively associated. Discussion Mental health sections (MHS) a t the 21 institutions studied in this report were found to have widely varying usage rates. The results of this study indicate that a very substantial proportion of this variability (95%) can be accounted for by five variables drawn from a pool o f 14 predictors. That is, by modifying four or five variables (such as the availability o f offcampus mental health resources, staffing levels and patterns, J. A. C. H. A.
USAGE R A T E
Table 3 REGRESSION A N A L Y S I S E M P L O Y I N G T H E F I V E V A R I A B L E S T H A T BEST ACCOUNT FOR USAGE R A T E V A R I A N C E Unstandardized Regression Coefficients Variable1
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
Weekly Interview Hours Per 1 0 0 0 FTS Percent o f Personnel T h a t Are Full-Time Number o f Visits Per Out-Patient Percent o f Service T i m e Devoted t o Consultative Service Availability of Off-Campus Mental Health Services (l=negligible, 2=moderate, 3=ample) Regression Equation Constant Term F o r Regression Using A l l Five Predictors
Percent o f T o t a l Usage Rate Variance Accounted F o r
Std. Error
F ratio2
Unique t o Each Variable2
3.9017 0.6012 -7.0142 -1.7203
0.5147 0.0799 1.1932 0.391 2
86.1 9 84.83 51.83 29.01
29.1 28.6a 17.51a 9.7
19.2890
4.81 7 6
24.05
8.1
94.9
37.54
93.0
94.9
Value
Cumulative With E n t r y 29.2 57.0 72.9 86.8
2.7653
1: Variables are listed in order o f entry i n t o the stepwise regression. 2: F o r each variable, degrees o f freedom are 1 and 15. F o r 5 variable regression equation, degrees o f freedom are 5 and 15. A l l f values are significant at the .001 level. a: Values shown include extensive supressor effects.
or the allocation o f staff time) and thereby making institutions more nearly alike, one could expect their usage rates to be more nearly alike. For some o f these variables, the results indicated that changing their values would affect usage rate in common sensical and unsurprising ways. For example, usage rate can be expected t o increase as the proportion o f on-campus resources represented by the MHS i s increased. Apparently, the larger the entrance ~- relative to other doorways - the heavier the traffic. Ten years ago, Reifler and Liptzin noted that increased professional service time (staffing level) was thought to be one factor accounting for reported increases in usage rate over time.7 The results o f this study both support that conjecture and provide an estimate o f the importance o f staffing level (i.e., 29% of usage rate variance). Usage rate was also found to be positively associated with the rate o f leaves for mental health reasons. A causal association i s unlikely here. Rather, both variables may be indices o f the balance o f stress and support experienced by students. When stress i s high, both indices tend t o increase. A positive association was also found between usage rate and the proportion o f students living in university housing. This may reflect primarily the ratio o f undergraduates t o graduate students. Residential proximity to the mental health section would obviously lower circumstantial barriers to using the MHS, and undergraduates are more likely to choose (or be required) t o live in university housing than graduate students. Thus, increasing the proportion o f students living in residence halls can be expected t o increase usage rate. Changes in other variables would apparently affect usage rate in unexpected ways. Usage rates were found to be higher, for example, when off-campus resources were more readily available and when a medical school existed on campus. These two counterintuitive results suggest that the presence o f ample off-campus resources and o f a medical school on-campus may contribute to a climate o f awareness and acceptance o f oncampus resources, lowering attitudinal and informational barriers to student use o f the mental health section. For a third group o f variables, the e f f e c t of changes in their values on usage rate i s more provocative than either expected or paradoxical. One example is the positive association found between usage rate and the proportion o f professional personnel who are full-time and/or have psychology backgrounds. These effects may arise because such personnel tend to be more involved in research. Greater staff involvement with VOL. 28, DECEMBER 7979
research was associated with a higher usage rate, suggesting that such activity, which often involves contact with students in the student-teacher role, may serve a case finding function and/or reduce the psychological barriers t o use. (Staff participation in formal academic instruction, though not a statistically significant predictor, was also positively related t o usage rate, possibly for the same reason.) The Mental Health Annual Program Survey, the source o f the data on which this report is based, has had as one o f i t s goals the development o f comparative data covering a variety o f areas related to mental health section programs and practices. Professionals in the field o f college mental health have questioned the utility and validity o f such comparative data, pointing to the obvious and more subtle differences between programs and their institutional settings. The analyses reported here represent both a validation o f these concerns and an attempt t o address and surmount them. Validation o f these concerns is provided by r-esults indicating that if the mental health sections a t different institutions have different usage rates, these differences can be almost completely accounted for by a relatively small number o f factors. Some o f these, like the availability o f off-campus mental health resources or the presence o f a medical school on-campus, are largely or completely outside the control o f college mental health administrators. Others, like staffing patterns and the allocation o f s t a f f time, would appear to be more amenable to modification. Both kinds of factors, however, can be taken into consideration using the regression techniques reported here. A comparison o f appropriately adjusted usage rates could reveal differences in levels o f student stress a t different institutions or the impact o f different programming strategies. I s a higher (or lower) usage rate a goal to be pursued? Such questions lie outside the scope o f this report. However, once a goal has been established, the analyses reported can both inform decisions designed t o implement these goals and increase understanding o f the impact o f unmodifiable factors. While care has been taken in these analyses to apply standards o f statistical significance in an appropriate and consistent manner, many o f the familiar cautions apply to interpretation or application o f these results. The sample size is relatively small. Regression analyses capitalize on random error, and validation o f the predictive equation on a second sample o f data i s required to define the extent t o which this has occurred here. Concerns regarding the use o f a large initial pool o f 143
COLLEGE H E A L T H
predictors i s only partially mitigated by the results o f preliminary analyses employing smaller subsets o f variables. Finally, the operation o f redundancy and suppressor effects underlines the complexity o f the relationship both among the predictors and also between usage rate and these predictors. Perhaps the most appropriate stance is to regard these analyses as a useful first step toward increasing our understanding of usage rates for college mental health programs and as an example o f one approach t o a variety o f questions about this field. REFERENCES
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
1. Reifler CB, e t al: A student mental health data bank: report o f
a pilot project. J Amer Coll Health Assoc 21:405-406, 1973 2. /bid 3. Greene EF, Schwartz AJ: Summary Report: Mental Health Annual Program Survey, 1973-74. Evanston, Illinois: American College Health Association, 1975 4. Schwartz A J : Mental Health Annual Program Survey (MHAPS): Summary Report for the 1975-76 Operating Year. Evanston, Illinois: American College Health Association, 1978 5. Reifler CB, Liptzin MD: Epidemiological Studies of College Mental Health. Arch Gen Psychiat 20:528-540, 1969 6. Nie NH, Hull CH, Jenkins JG, Steinbrenner K and Bent DH: SPSS: Statistical Package for the Social Sciences. New York: McGrawHill, 1975 7. Reitier CB, Liptzin MD: op cit
144
J. A .C. H. A.
Journal of American College Health Association Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/vzch20
Factors Affecting the Usage Rate for Mental Health Sections of College and University Health Services Allan J. Schwartz Ph.D. a
a b
University of Rochester, School of Medicine and Dentistry , USA
b
University Health Service, University of Rochester , 250 Crittenden Blvd., Box 617, Rochester, New York, 14642, USA Published online: 07 Apr 2011.
To cite this article: Allan J. Schwartz Ph.D. (1979) Factors Affecting the Usage Rate for Mental Health Sections of College and University Health Services, Journal of American College Health Association, 28:3, 140-144, DOI: 10.1080/01644300.1979.10392917 To link to this article: http://dx.doi.org/10.1080/01644300.1979.10392917
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Factors Affecting the Usage Rate for Mental Health Sections of College and University Health Services ALLAN J . SCHWARTZ, Ph.D.* University o f Rochester
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
Abstract The history o f the Mental Health Annual Program Survey (MHAPS) i s briefly reviewed. Characteristics o f the institutions participating in MHAPS during the 1975-76 program year are reported. Usage rate data for the sample are reviewed, and the procedures employed in determining the predictors o f sample variability in mental health section usage rate are described. Variability in usage rate was regressed against six categories o f variables: institutional characteristics, alternative available resources, mental health section staffing levels and staffing patterns, allocations o f staff time, and programming options. When the significant predictors included in these six categories were pooled, five variables were sufficient t o account for 95% o f usage rate variance. The number o f weekly interview hours per 1000 full-time students and the proportion o f full-time staff each accounted for 29% o f usage rate variance. Visits per patient accounted for 18%, time allocated t o consultative services for lo%, and the availability o f off-campus resources for 8%. The effects that changes in the values o f the predictors can be expected t o have on usage rate are illustrated, and the implications o f these results for the use o f comparative data are discussed. Introduction In January 1972, a t a conference in Boulder, Colorado sponsored by the Western Interstate Commission on Higher Education, a decision was made to establish a systematic, annual survey o f the operations o f the mental health sections o f college and university health services.’ Originally named the Student Mental Health Data Bank, the program was renamed the Mental Health Annual Program Survey (MHAPS) in 1973. Since i t s inception, MHAPS has been conducted under the sponsorship o f the American College Health Association. The substantive areas covered by the survey include amount and modes o f service, staffing levels and patterns, complementary formal mental health programs, suicide, official leaves o f absence due t o illness, and allocation o f professional time. Selected data from the first survey, covering the 1970-71 operating year, have been published. Unpublished summaries reviewing data for the 1973-74 and 1975-76 operating years have been distributed t o survey respondents through the American College Health A~sociation.3~4 This report focuses on usage rate data from the 1975-76 survey.
Modally, the sample consisted o f publicly funded (61%), doctorate-awarding (74%) universities (83%) with a median full-time enrollment o f 14,000. The range o f enrollment was quite large, however, with enrollments as small as 1700 and as large as 45,000 being represented. The median proportion o f students living in universityowned or operated housing was 37% o f full-time student enrollment. Sixty-one percent o f the institutions had two or more on-campus agencies that complemented the mental health section, while only 13% had no other on-campus resource for students. The most usual complementary agency was a counseling center. Off-campus mental health resources were most often designated as ample (44%) or moderate (35%), and 39% reported having a medical school on-campus. The geographic distribution o f the sample showed 61% o f the institutions located in New England and in other states bordering the Atlantic. Only one institution was located in the Rocky Mountain and Pacific Coast states, and 35% were located in states lying between the Appalachian and Rocky Mountains.
The Sample o f Respondents
Usage Rate
Twenty-eight institutions responded t o requests for survey data for the 1975-76 operating year, with five o f those reporting that they did not have a designated mental health staff. Based upon past surveys, it i s estimated that about 100 o f the 400 institutions solicited had a designated mental health staff. Accordingly, the sample o f 23 respondents upon which this report i s based represents a return rate o f 20 to 25%.
Usage rate, defined as the number o f different patients seen for each 1000 full-time students enrolled per 12 month operating year (pts/l OOOfts), was derived from patient and enrollment data provided by the reporting institutions. The enrollment figures supplied were for the fall term (semester or quarter) o f the 1975-76 operating year. The 21 institutions providing both patient and enrollment data represented 309,413 full-time students and provided services to 13,753 different patients. For the sample as a whole, therefore, the usage rate was 44pts/l000fts. There was notable variability in usage rate across institutions. The mean o f the 21 individual usage rates, 54pts/lOOOfts, was
*Assistant Professor of Psychiatry (Psychology), University of Rochester School of Medicine and Dentistry, and Staff Psychologist, University Health Service, University of Rochester, 250 Crittenden Blvd., B o x 617, Rochester, New Y o r k 14642
140
J.A.C.H.A.
USAGE R A T E
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
somewhat higher than the median o f 40pts/l OOOfts, largely as a function o f relatively high usage rates a t two institutions with relatively small enrollments (1 7OOfts and 6900fts). Usage rates for these institutions were 179 and 115 respectively, values well above the range within which the remainder o f the sample fell (10-90pts/lOOOfts). The mean, median and range o f usage rates for this sample are quite consistent with those reported in previous studies o f usage rates for college mental health sections.5 The wide variability in usage rate among this sample o f institutions stimulated a search for the correlates and possible determinants o f usage rate. Six categories of variables that seemed likely t o influence usage rates were selected. These categories and the variables initially included in each are listed in Table 1. The decision t o include a particular variable in this preliminary phase was based upon a variety o f criteria including that variables’s intrinsic interest as a predictor, i t s distributional characteristics, and the absence of missing data.
Method Using the statistical package for the Social Sciences (SPSS) programming package, each category o f variables was entered into a separate stepwise multiple regression analysis.6 This permitted identification o f the most parsimonious subset o f statistically significant predictors within each category. Additional analyses were then done using only those variables that had been statistically significant predictors in the preceding analyses. For each category, each variable was entered as the last predictor o f the set t o determine the unique usage rate variance accounted for by that variable. Then the common variance, that is, the proportion o f usage rate variance accounted for by either o f two (or more) variables in each s e t o f predictors was assessed.? In a final set o f analyses, predictors from all six categories were entered into the regression analyses t o determine the single best subset of predictors and t o determine the proportions o f unique and o f common variance accounted for by these variables. Results
Table 1 PREDICTOR V A R I A B L E S USED IN T H E REGRESSION A N A L Y S E S C A T E G O R Y 1: MOSAIC OF A V A I L A B L E RESOURCES Availability o f Off-Campus Services Medical School o n Campus Number o f Special Formal Programs Number o f Other On-Campus Agencies Providing Related Services Percent o f T o t a l On-Campus Weekly Interview Hours Represented By Mental Health Section Interview Hours C A T E G O R Y 2: I N S T I T U T I O N A L C H A R A C T E R I S T I C S Size (full-time enrollment) Institutional Designation (college vs. university) Funding (public vs. private) Percentage o f Students Living O n Campus Highest Degree Awarded Rate o f Official Leaves o f Absence f o r Mental Health Reasons C A T E G O R Y 3: A L L O C A T I O N OF PROFESSIONAL E F F O R T ( T I M E ) Percent in Direct Services Percent in Indirect Services Percent in Consultative Services Percent i n Formal Academic Instruction Percent in Providing Clinical Supervision Percent i n Receipt o f In-Service Training Percent in Doing Research C A T E G O R Y 4: S T A F F I N G PATTERNS Percent F u II-Time Percent w i t h Staff Status (vs. trainees) Percent w i t h Psychiatry Background Percent w i t h Psychology Background Percent w i t h Social Work Background Percent w i t h Nursing Background CATEGORY 5: STAFFING L E V E L Number o f Full-Time Equivalent Personnel Per 1000 Full-Time Students Number o f Weekly Interview Hours Per 1000 Full-Time Students C A T E G O R Y 6: P R O G R A M M I N G OPTIONS Percent o f T o t a l Services = Direct Services Percent o f Outpatient Services = Group Services Percent o f Outpatient Services = Couples/Family Services Visits Per Patient Percent o f Patients Referred f o r Further Care
VOL. 28, DECEMBER 1979
Each category o f variables except programming options was found t o account for a statistically significant proportion o f usage rate variance. As shown in Table 2, the category accounting for the greatest proportion o f usage rate variance, 7376, was the set o f variables that described the mosaic of mental health resources available to students. This result suggests that the presence of alternative resources (both on and off-campus), the existence o f specialized formal programs (such as pregnancy counseling, crisis phone services, and courses in human sexuality), and the proportion o f on-campus resources represented by the mental health section all affect usage rate. Mental health section usage rate was lower when the total number o f specialized formal programs was high, and usage rate was higher when there was a medical school on campus, when off-campus resources were more readily available, and when the mental health section represented a higher proportion o f on-campus resources. Fifty-seven percent o f usage rate variance was accounted for by the second ran king cateogry, institutional characteristics. Offical leaves o f absence for mental health reasons was the single best predictor in this category. This variable alone accounted for 34% o f usage rate variance with a higher rate of leaves predictive o f higher usage rate. When the proportion of students living in university-owned or operated housing was high, so was usage rate. The final significant predictor in this category, highest degree awarded, was negatively related t o usage rate. That is, institutions awarding doctorates tended t o have lower usage rates than those awarding master’s degrees. Allocation o f staff effort (time), the third ranking category, accounted for 53% of usage rate variance. The allocation o f a larger proportion o f staff time to indirect and consultative services was associated with lower usage rate. Greater staff involvement with research was associated with a higher usage rate. In accounting for statistically significant portions o f usage rate variance, staffing patterns was the fourth ranked category (32%), with staffing level fifth and last (29%). TWO staffing pattern parameters, the proportion o f staff designated
tSuppressor effects were also noted f o r variables i n five o f the six categories and f o r the pooled s e t o f variables. These effects indicate t h a t the value o f one predictor ( X i ) is affecting a second predictor ( X 2 ) w i t h the result t h a t removing the effect o f X i on X 2 changes the correlation between X 2 and Y. More technically, the zero order correlat i o n [ryx2] is n o t equal t o the semipartial correlation [ r y ( x 2 . x 1 ) ] .
141
COLLEGE HEALTH
Table 2 USAGE RATE VARIANCE ACCOUNTED FOR BY VARIABLES I N EACH O F SIX CATEGORIES Percent Of Total Usage Rate Variance Accounted For1
Direction of Association
1. MOSAIC OF A V A I L A B L E RESOURCES Availability of Off-Campus Services Medical School On Campus (l=yes, 2=no) Number o f Special Formal Programs Percent o f Total OnCampus Weekly Interview Hours Represented by Mental Health Section
73 41 a 7a 9a 35a
positive negative negative Dositive
2. IN ST ITUTION A L CH A RACTE R ISTl CS Percent o f Full-Time Students Eligible for Mental Health Services Rate o f Official Leaves o f Absence for Mental Health Reasons Highest Degree Awarded
57 18a
positive
.05
34
positive
.o 1
10
negative
.01
3. ALLOCATION OF PROFESSIONAL EFFORT (TIME) Percent o f A l l Service Time in Indirect Services Percent o f A l l Service Time in Consultative Services Percent o f Total Time Doing Research
36 6 8 10
negative negative positive
.05 .05 .05
4. STAFFING PATTERNS (base = total full time equivalents of professional personnel) Percent Fu II-T ime Percent with Psychology Background
32 11 14
positive positive
.05 .05
5. STAFFING LEVEL Number of Weekly Interview Hours Per 1000 Full-Time Students
29 29
positive
.05 .05
6. PROGRAMMING OPTIONS Visits Per Out-Patient Percent o f Patients Referred for Further Care
26 14a 13a
negative negative
.10 .10 .10
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
Category and Variables
Significance Level
.o 1 .o 1 .01
.o 1 .01
.o 1
.01
.05
1: Values for categories = square o f multiple correlation. a: Percentage o f variance accounted for reflects supressor effects.
full-time and the proportion with psychology training (vs. psychiatry, social work, nursing) were positively and significantly related to usage rate. In terms o f staffing level, both the number o f full-time equivalent staff per 1000 full-time students (fts) and the number o f weekly interview hours per 1000fts were positively related to usage rate. It i s apparent from summing the proportions o f usage rate variance accounted for by each category o f variables that sizable portions of the variance accounted for by variables in one category would also be accounted for by variables in other categories. This was confirmed when categories were pooled and the s e t o f twelve previously significant variables, augmented by two nearly significant programming option variables, were entered into a stepwise regression analysis. Rate o f mental health leaves, proportion o f personnel with psychology background, proportion o f time spent in research, visits per patient, proportion of all service time devoted t o consultative service, weekly interview hours per 1OOOfts, and proportion o f personnel with full-time commitment were the predictors selected. However, when the unique contribution o f each variable was assessed in relation t o the remaining six, rate o f leaves and time in research were found t o make unique contributions o f less than 1%. This provided both dramatic and specific confirmation o f the redundancy that had been anticipated. Rate o f leaves and time spent in research were removed from the set o f 14 predictors, and a second s e t o f regression analyses was performed. Table 3 summarkes the results o f 142
these analyses. Five variables accounting for 95% o f usage rate variance now entered the regression equation. These were weekly interview hours per 1OOOfts, proportion o f personnel with full-time commitment, number o f visits per outpatient, proportion o f service time devoted to consultative services, and the availability o f off-campus mental health services. Each o f these variables provided a unique and statistically significant increment in usage rate accounted for, within the final set o f five predictors. The stuffing level and stuffing pattern variables (weekly interview hours per 1OOOfts and proportion o f full-time personnel) each accounted for 29% o f usage rate variance. Visits per out-patient accounted for 17%, proportion o f time devoted to consultative services for lo%, and availability o f off-campus resources for 8%. Only 2% o f the composite 95% o f usage rate variance accounted for was common or overlapping variance. Visits per outpatient and proportion o f service time devoted to consultative services were negatively associated with usage rate, while the other three variables were positively associated. Discussion Mental health sections (MHS) a t the 21 institutions studied in this report were found to have widely varying usage rates. The results of this study indicate that a very substantial proportion of this variability (95%) can be accounted for by five variables drawn from a pool o f 14 predictors. That is, by modifying four or five variables (such as the availability o f offcampus mental health resources, staffing levels and patterns, J. A. C. H. A.
USAGE R A T E
Table 3 REGRESSION A N A L Y S I S E M P L O Y I N G T H E F I V E V A R I A B L E S T H A T BEST ACCOUNT FOR USAGE R A T E V A R I A N C E Unstandardized Regression Coefficients Variable1
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
Weekly Interview Hours Per 1 0 0 0 FTS Percent o f Personnel T h a t Are Full-Time Number o f Visits Per Out-Patient Percent o f Service T i m e Devoted t o Consultative Service Availability of Off-Campus Mental Health Services (l=negligible, 2=moderate, 3=ample) Regression Equation Constant Term F o r Regression Using A l l Five Predictors
Percent o f T o t a l Usage Rate Variance Accounted F o r
Std. Error
F ratio2
Unique t o Each Variable2
3.9017 0.6012 -7.0142 -1.7203
0.5147 0.0799 1.1932 0.391 2
86.1 9 84.83 51.83 29.01
29.1 28.6a 17.51a 9.7
19.2890
4.81 7 6
24.05
8.1
94.9
37.54
93.0
94.9
Value
Cumulative With E n t r y 29.2 57.0 72.9 86.8
2.7653
1: Variables are listed in order o f entry i n t o the stepwise regression. 2: F o r each variable, degrees o f freedom are 1 and 15. F o r 5 variable regression equation, degrees o f freedom are 5 and 15. A l l f values are significant at the .001 level. a: Values shown include extensive supressor effects.
or the allocation o f staff time) and thereby making institutions more nearly alike, one could expect their usage rates to be more nearly alike. For some o f these variables, the results indicated that changing their values would affect usage rate in common sensical and unsurprising ways. For example, usage rate can be expected t o increase as the proportion o f on-campus resources represented by the MHS i s increased. Apparently, the larger the entrance ~- relative to other doorways - the heavier the traffic. Ten years ago, Reifler and Liptzin noted that increased professional service time (staffing level) was thought to be one factor accounting for reported increases in usage rate over time.7 The results o f this study both support that conjecture and provide an estimate o f the importance o f staffing level (i.e., 29% of usage rate variance). Usage rate was also found to be positively associated with the rate o f leaves for mental health reasons. A causal association i s unlikely here. Rather, both variables may be indices o f the balance o f stress and support experienced by students. When stress i s high, both indices tend t o increase. A positive association was also found between usage rate and the proportion o f students living in university housing. This may reflect primarily the ratio o f undergraduates t o graduate students. Residential proximity to the mental health section would obviously lower circumstantial barriers to using the MHS, and undergraduates are more likely to choose (or be required) t o live in university housing than graduate students. Thus, increasing the proportion o f students living in residence halls can be expected t o increase usage rate. Changes in other variables would apparently affect usage rate in unexpected ways. Usage rates were found to be higher, for example, when off-campus resources were more readily available and when a medical school existed on campus. These two counterintuitive results suggest that the presence o f ample off-campus resources and o f a medical school on-campus may contribute to a climate o f awareness and acceptance o f oncampus resources, lowering attitudinal and informational barriers to student use o f the mental health section. For a third group o f variables, the e f f e c t of changes in their values on usage rate i s more provocative than either expected or paradoxical. One example is the positive association found between usage rate and the proportion o f professional personnel who are full-time and/or have psychology backgrounds. These effects may arise because such personnel tend to be more involved in research. Greater staff involvement with VOL. 28, DECEMBER 7979
research was associated with a higher usage rate, suggesting that such activity, which often involves contact with students in the student-teacher role, may serve a case finding function and/or reduce the psychological barriers t o use. (Staff participation in formal academic instruction, though not a statistically significant predictor, was also positively related t o usage rate, possibly for the same reason.) The Mental Health Annual Program Survey, the source o f the data on which this report is based, has had as one o f i t s goals the development o f comparative data covering a variety o f areas related to mental health section programs and practices. Professionals in the field o f college mental health have questioned the utility and validity o f such comparative data, pointing to the obvious and more subtle differences between programs and their institutional settings. The analyses reported here represent both a validation o f these concerns and an attempt t o address and surmount them. Validation o f these concerns is provided by r-esults indicating that if the mental health sections a t different institutions have different usage rates, these differences can be almost completely accounted for by a relatively small number o f factors. Some o f these, like the availability o f off-campus mental health resources or the presence o f a medical school on-campus, are largely or completely outside the control o f college mental health administrators. Others, like staffing patterns and the allocation o f s t a f f time, would appear to be more amenable to modification. Both kinds of factors, however, can be taken into consideration using the regression techniques reported here. A comparison o f appropriately adjusted usage rates could reveal differences in levels o f student stress a t different institutions or the impact o f different programming strategies. I s a higher (or lower) usage rate a goal to be pursued? Such questions lie outside the scope o f this report. However, once a goal has been established, the analyses reported can both inform decisions designed t o implement these goals and increase understanding o f the impact o f unmodifiable factors. While care has been taken in these analyses to apply standards o f statistical significance in an appropriate and consistent manner, many o f the familiar cautions apply to interpretation or application o f these results. The sample size is relatively small. Regression analyses capitalize on random error, and validation o f the predictive equation on a second sample o f data i s required to define the extent t o which this has occurred here. Concerns regarding the use o f a large initial pool o f 143
COLLEGE H E A L T H
predictors i s only partially mitigated by the results o f preliminary analyses employing smaller subsets o f variables. Finally, the operation o f redundancy and suppressor effects underlines the complexity o f the relationship both among the predictors and also between usage rate and these predictors. Perhaps the most appropriate stance is to regard these analyses as a useful first step toward increasing our understanding of usage rates for college mental health programs and as an example o f one approach t o a variety o f questions about this field. REFERENCES
Downloaded by [Indiana University Libraries] at 22:30 16 March 2015
1. Reifler CB, e t al: A student mental health data bank: report o f
a pilot project. J Amer Coll Health Assoc 21:405-406, 1973 2. /bid 3. Greene EF, Schwartz AJ: Summary Report: Mental Health Annual Program Survey, 1973-74. Evanston, Illinois: American College Health Association, 1975 4. Schwartz A J : Mental Health Annual Program Survey (MHAPS): Summary Report for the 1975-76 Operating Year. Evanston, Illinois: American College Health Association, 1978 5. Reifler CB, Liptzin MD: Epidemiological Studies of College Mental Health. Arch Gen Psychiat 20:528-540, 1969 6. Nie NH, Hull CH, Jenkins JG, Steinbrenner K and Bent DH: SPSS: Statistical Package for the Social Sciences. New York: McGrawHill, 1975 7. Reitier CB, Liptzin MD: op cit
144
J. A .C. H. A.
