p,sychological and Social Correlates of Glycemic ... - Diabetes Care

p,

sychological and Social Correlates of Glycemic Control

ROGER S. MAZZE, Ph.D., DAVID LUCIDO, Ph.D., AND HARRY SHAMOON, M.D.

Eighty-four persons with insulin-dependent diabetes participated in this study to determine whether glycemic control was related to personality, anxiety, depression, and/or quality of life. The subjects were placed on either a conventional treatment regimen consisting of one to two injections of mixed short- and intermediate-acting insulin, with urine testing or an intensive treatment regimen consisting of two or more injections of mixed insulins, with self-monitoring of blood glucose. Personality was found to have no relationship to level of glycemic control either at the beginning of the study or at any point during the study. In contrast, anxiety, depression, and quality of life showed a significant relationship to metabolic control at entry and throughout the study period. Lower anxiety and depression scores and better quality of life scores were recorded for those subjects in good control (HbA! < 8.9%) when compared with those in average control (HbAj 9.0-11.9%) and those in poor control (HbA! > U-9%) at entry (P = 0.01). At each point during the study the difference between those in good control and those in poor control in terms of anxiety, depression, and quality of life was significant (P = 0.02). Change in glycemic control was found to account for up to 20% of the betweenpatient variability for these psychosocial parameters, DIABETES CARE 1984; 7:360-66.

D

iabetes, like other chronic diseases, may have a major impact on the individual's psychosocial status.1"6 Unlike other chronic diseases, however, the extraordinary effort required for selfcare as well as the ever-present possibility of developing complications may present added psychological and social stresses that would adversely affect the person with diabetes. Previous studies, which have examined the effect of diabetes on psychological and social characteristics, are divided as to whether persons with diabetes differ significantly from persons without diabetes.7 Few studies have attempted to evaluate the relationship between psychosocial factors and glycemic control, due largely to the absence of adequate measurements of longterm metabolic control. While earlier studies have examined the psychological changes that may occur with and following the diagnosis of diabetes8"10 as well as the psychological differences between persons with diabetes and persons without diabetes,u"13 few investigators14 have examined such dynamic psychosocial characteristics as quality of life, depression, and anxiety as they relate to glycemic control. These latter issues are of particular concern in view of the current interest in more stringent metabolic control and the prevention of com-

360

plications.1516 We therefore set out to examine: (1) whether there are psychological and social characteristics unique to persons with insulin-dependent diabetes mellitus (IDDM); (2) whether such characteristics are related to the degree of glycemic control; and (3) whether different treatment approaches are associated with changes in these psychosocial variables. MATERIALS AND METHODS

Subjects. One hundred fifteen individuals with IDDM responded to advertisements in local and regional newspapers. Telephone screening for age (13-41 yr), absence of debilitating complications, no current pregnancy, and accessibility to the diabetes study site reduced the initial subject pool to 84. This subject pool consisted of 59 women and 25 men with an average age of 26.9 yr (±0.8), and an average duration of diabetes of 11.1 yr (±0.9). This predominantly urban population (70%) was 60% Caucasian, 22% black, and 18% Hispanic. Fifty-five percent of the subjects were employed, 15% were unemployed, 12% were homemakers, and 18% were attending school. The study group was com-

DIABETES CARE, VOL. 7 NO. 4, JULY-AUGUST 1984

PSYCHOLOGICAL AND SOCIAL CORRELATES OF GLYCEMIC CONTROL/R. S. MAZZE, D. LUC1DO, AND H. SHAMOON

infusion (using the Auto-Syringe pump model AS6C, AutoSyringe Inc., Hooksett, New Hampshire). Measurements. Glycemic control was assessed at 6-wk inRandomized Nonrandomized tervals using HbA].17 Due to temperature and interassay varsubjects subjects iability of the resin-binding column method, results are reVariable list (N = 50) (N = 34) ported together with a control value from a normal, nondiabetic Sex 73% (F)/27% (M) 73% (F)/27% (M) red cell hemolysate. The difference between the percent glyAge 25.5 ± 1.1 27.7 ± l.Ot cosylated hemoglobin in the patient and that in the control Total HbA, (%) 10.8 ± 0.4 10.2 ± 0.4 is expressed as a delta HbA^ (delta 4.5 ± 0.4) (delta 4.2 ± 0.5) Trait- (stable) and state- (dynamic) dependent self-admin52.9 ± 2.5 Total HDL (mg/dl) 45.4 ± 2.1 istered psychological measures were completed at entry and Plasma triglycerides at 18-wk intervals. The Emotions Profile Index was used to 103.6 ± 't•2.1 (mg/dl) 90.1 ± 9.3 characterize four trait-dependent personality dimensions: timid/ Plasma cholesterol aggressive, trustful/distrustful, controlled/dyscontrolled, and (mg/dl) 176.5 ± 5.7 168.5 ± 7.1 18 gregarious/depressed. The Taylor Manifest Anxiety Scale19 Knowledge 16.9 ± 0.7 14.7 ± 0.8 and the Zung Self-Rated Depression Scale20 were used to Anxiety 3.0 ± 0.5 3.1 ± 0.3 gauge state-dependent psychological variables. These scales, Depression 4.3 ± 0.7 5.7 ± 0.6 13.6 ± 2.0 PCL 16.7 ± 2.0 modified by the deletion of items related to physical symp60/30/10 60/30/10 *SES (WC/MC/UC) (% ) toms, use self-descriptive statements to evaluate the current level of anxiety and depression, respectively. They are sen* Socioeconomic status: WC = working class; MC = middle class; UC sitive to transient shifts in these variables among chronically upper class. ill individuals. t All data except sex and SES are given as X ± SE. For evaluation of quality of life, the Mooney Problems Check List (PCL) was adapted (by omission of a section specifically related to college students) for persons with diposed of approximately 60% lower/working class, 30% midabetes.21 This self-administered protocol is composed of 100 dle class, and 10% upper middle class. brief descriptions of problems, which are grouped into six Procedures. Subjects were to be randomized onto one of major categories: perceptions about health, job-related probtwo principal treatment modalities: (1) "conventional" therlems, personal problems, interpersonal problems, family apy, using one to two injections of insulin in varying mixproblems, and sexual relations. The scale provides an evaltures, and urine glucose testing, or (2) "intensive" therapy, uation of the "amount of problems currently experienced in using two to four injections of mixed insulins per day, and each of the areas." Additionally, by summing all of the probcapillary blood testing. Urine fractionals were to be perlems checked in each category, a total life problems measure formed on single- or double-voided specimens four times per can be derived. The benefit of this scale is that it provides day using the two-drop Clinitest method (Ames Division, a current measure of the "stresses and conflicts" related to Miles Laboratories, Elkhart, Indiana). Capillary blood testing daily living. All psychosocial tests were administered before was to be performed with Dextrostix (Ames) and either the the subjects' contact with the health care providers. This Dextrometer (Ames) or Glucometer (Ames) reflectance meinsured that feedback from the providers did not directly ters. Subjects were asked to record insulin injections, tests, impact on the subjects' responses on these protocols. diet, exercise, and any symptoms of hyper- or hypoglycemia in daily logbooks. Visits to the diabetes unit were set at 6wk intervals. TABLE 2 Volunteers attended an orientation program, which de- 18th and 36th week data for randomized and nonrandomized subjects tailed the different treatment modalities and the randomiRandomized Nonrandomized zation procedures. Fifty subjects gave informed consent to be randomized. Thirty-four volunteers declined to be random- 18th week (N = 25) (N = 41) ized because of a stated preference for one of the treatment HbA, delta 3.6 ± 0.3* 3.6 ± 0.4 regimens. Both groups consented to participate in all other Anxiety 1.6 ± 0.3 2.0 ± 0.5 Depression 3.9 ± 0.7 3.4 ± 0.8 study procedures: attending the clinic every 6 wk, laboratory PCL 9.8 ± 2.4 8.4 ± 2.1 tests, self-administered social and psychological tests, main(N = 20) (N = 34) tenance of a logbook, and willingness to meet the goal of 36th week HbA, delta 4.2 ± 0.7 3.7 ± 0.4 improved glycemic control. Due to sequential entry over a 1.2 ± 0 . 2 Anxiety 2.7 ± 0 . 7 4-mo period, subjects were individually randomized resulting Depression 2.9 ± 0.5 3.3 ± 0.9 in 28 patients in the intensive treatment, and 22 in the PCL 9.0 ± 2.0 13.1 ± 3 . 0 conventional treatment groups. Of the 34 nonrandomized subjects, 28 selected intensive treatment, 4 chose conven- 'Standard error. These values were not significantly different for the two tional therapy, and 2 chose continuous subcutaneous insulin groups. TABLE 1 Entry data


361

PSYCHOLOGICAL AND SOCIAL CORRELATES OF GLYCEMIC CONTROL/R. S. MAZZE, D. LUCIDO, AND H. SHAMOON

TABLE 3 Problems Check List at entry

two groups. Furthermore, analysis of physiologic (glycemic control), social (quality of life), psychological (personality, anxiety, depression), and performance (logbook) data (Table PCL Categories SE 2) for each of these groups during the study period revealed no significant differences. Health 0.35 4.23: The results of the personality profile (EPI) performed at 0.18 1.30: Vocational 0.90 7.60: Personal entry showed no significant difference between the scores for 0.24 1.70: Interpersonal the subjects and the normal scores (20-80) for this scale.18 0.20 1.06: Family Similarly, no significant difference was found for the anxiety 0.14 0.65: Sexual and depression scores of our subjects when compared with 15.63: 1.48 Total" standard scores. Mean anxiety scores of 3.1 ± 0.28 (normal range 0—8) and mean depression scores of 5.1 ± 0.45 (nor'Not significant when compared with normal mean score (18 ± 2.0). mal range 0-12) were found to be relatively low for both scales and thus nondiagnostic for underlying anxiety or depression. Statistical procedures. Data were entered into an Apple II Table 3 shows the mean scores for the six dimensions of computer with a data base system and electronically uploaded the Problems Check List (PCL) as well as the total score for into a DEC VAX 11/780 for evaluation using the Statistical 22 all subjects at entry. The total score represents an overall Analysis System. Non-pairwise t tests were used to determine whether a significant difference existed between our measure of quality of life, with 0 indicative of the absence sample and appropriate norms on each psychological and of problems, and 100 representing a high degree of problems social measure. For comparisons between groups at single of daily living. We could find no difference between the timepoints, an overall analysis of variance was used, followed scores of persons with diabetes and PCL normal scores. We then determined whether the degree of metabolic conby single degree of freedom contrasts for pairwise comparisons between groups divided according to their levels of metabolic trol was related to psychological and/or social characteristics control. Pearson correlation coefficients were computed for of the person with diabetes. Figure 1 depicts the distribution evaluation of the relationship between parameters across groups of delta HbA t values at entry for all subjects. The subjects clustered in a near-normal distribution with the HbAj values at multiple timepoints. ranging from 0% to 15% above the control values (i.e., delta HbA^. Since we were interested in the differences between RESULTS subjects in near-normal glycemic control and those in poorer his report is based on data collected from the ran- metabolic control, we divided the subjects into quartiles. domized (N = 50) and nonrandomized (N = 34) The mean delta HbA] of the 21 subjects in the first quartile subjects. We chose to pool the data (Table 1) from was 1.7%. The mean delta HbA! value in the second and these two groups because at entry, demographic third quartiles was 4.3% for the 42 subjects. For the 21 (age, sex, race, and socioeconomic status), physiologic, psy- subjects in the fourth quartile, the mean delta HbA] was chological, and social characteristics as well as previous treat- 8.5%. For the purpose of identification we designated the ment, knowledge of diabetes, and willingness to be treated glycemic control in these groups as good, average, and poor, on the diabetes unit did not differ significantly between these respectively.

T

02,3 (Average)

Q4 (Poor)

Number of Subjects

4 -

2

362

3

4

5

6 7 8 9 10 II Delta HbA, (%)

12 13 14 15 16

FIG. I. Distribution of subjects by level of glycemic control The open area represents 21 subjects in good control; the cross-hatched area represents 42 subjects in average control; and the shaded area represents 21 subjects in poor control.



FIG. 2. Entry psychosocial profile of subjects in good, average, and poor control. Means for sub' jects in good control (open bar), average control (cross-hatched bar), and poor control (shaded bar) are shown for anxiety, depression, and quality of life. Ranges for the variables were: (1) delta HbA,, 0-16%; (2) anxiety, 0-10; (3) depression, 0-15; and (4) PCL, 0-100.

26 24 22 20 18 16 14 12 10 8 6 4 2 0

Total Problem Check List Score

Delta HbA, D, Entry)

We then examined the psychosocial status of the members of each metabolic control group. An EPI personality profile for each of the metabolic control groups was produced. No difference in personality emerged when the patients were divided according to level of glycemic control. Next, the relationship between glycemic control and anxiety and depression at entry was evaluated (Figure 2). A pattern of decreasing scores in association with lower delta HbA] was noted. A significant difference was found between the group in good control and the group in poor control for both anxiety (F = 5.08, 1,82 df, P = 0.02) and depression (F = 5.39, 1,82 df, P = 0.02). Similarly, a trend of decreasing scores for PCL (hence better quality of life) with improved metabolic control was uncovered. The mean score for the group in good control was 12.7, for the average group, 13.7, and for the poor group, 22.4. The difference between those in good control and those in poor control in relation to PCL was significant (F = 6.23, 1,82 df, P = 0.01). We then followed the subjects through their 18th- and 36th-week monitoring visits. Sixty-six of the original 84 subjects were tracked to the 18th week, and 54 of these 66 subjects were followed to the 36th week. A drop-out rate of 21% was noted for the period between the entry and 18th weeks, which reduced to 18% between the 18th and 36th weeks. Drop-out was due to three factors: (1) inconvenience of clinic hours; (2) dissatisfaction with treatment (equal among both treatment groups); and (3) dissatisfaction with results. The drop-outs, however, did not differ significantly from those who remained in the study in terms of previous treatment experience, metabolic control, or demographic or socioeconomic variables.

Anxiety Score

We could find no significant difference among the personality profiles of those in good, average, or poor control throughout the study. In contrast, we found a relationship between glycemic control and anxiety and depression as shown in Figure 3. The range of delta HbA, that defined each glycemic control group was set at entry. The data show that as glycemic control worsened, both anxiety and depression increased. At the 18th week, with a reduction in mean delta HbA, for the poorly controlled group—from 8.5% to 6.7%— there was a corresponding drop in the level of anxiety and depression. At the 36th week, the anxiety and depression scores increased for the poorly controlled group, as did the delta HbA,. This pattern was repeated in the measure of quality of life. Figure 3 shows that at the 18th week the fall in delta HbA, by nearly 2% corresponded with a fall in PCL by 10 points for those in poor control. At the 36th week the PCL increase corresponded with an increase in delta HbA,. To determine whether the subjects reaching the 18th and the 36th weeks differed significantly from the initial pool, we followed both groups back to entry. We found no significant difference within the metabolic control categories for any of the psychosocial variables (not shown). Could these findings be due to the relative stability of membership within each glycemic control group throughout the study period? By the 18th week the 66 subjects had redistributed themselves in terms of control: 26, good; 30, average; and 10, poor. Of the 26 patients in good control, 42% came from those originally in good control, 50% originally in average control, and 8% originally in poor control. This pattern of patient migration repeated itself at the 36th week. Overall, 60% of the patients changed metabolic groups by the 36th week.


363


Delta HbAi (%)

Anxiety Score

18th Week

36 th Week

FIG. 3. Follow-up psychosocial profile of subjects in good, average, and poor control. The open bar represents subjects in good control, the crosshatched bar represents subjects in average control, and the shaded bar represents subjects in poor control as defined by entry glycemic control distribution.

Next we compared the differences in physiologic and behavioral parameters for the period from entry to 36 wk. As illustrated in panel A of Figure 4, a significant correlation (r = 0.46, P = 0.001) was found between changes in HbA! level and changes in level of depression. As metabolic control worsened, the depression score increased. A similar finding (Figure 4, panel B) was uncovered with regard to anxiety.

Increased HbA! was correlated (r = 0.33, P = 0.03) with higher anxiety scores. Finally, as shown in panel C of Figure 4, quality of life, as measured by the Problems Check List, improved with better glycemic control (r = 0.31, P = 0.03). This single variable model (change in anxiety, depression, or quality of life, and change in glycemic control) accounted for up to 20% of the between-patient variability in these psychosocial parameters. To determine whether these results were confounded by age, sex, or socioeconomic characteristics of the subjects, linear regression analysis was performed. The relationship between glycemic control, anxiety, depression, and quality of life was examined, controlling for the contribution of the demographic parameters. This analysis showed that sex, age, and socioeconomic status did not independently contribute to the relationship between glycemic control and these psychosocial variables. Finally, to determine whether treatment type—conventional versus intensive—impacted on the dependent variables, we analyzed both the pooled data and data from the randomly assigned group alone. Analysis of variance revealed that treatment type did not significantly impact on metabolic control, anxiety, depression, or quality of life for the pooled data. To examine whether the subjects randomized to conventional therapy had a significantly different outcome than those randomized to intensive treatment, we compared metabolic control, anxiety, depression, and quality of life data at the 18th and 36th weeks of the study. Table 4 shows that there was no significant difference on any of these variables between the two groups. DISCUSSION

W

e found that persons with type I diabetes do not manifest any significant differences in personality, anxiety, depression, and quality of life when compared with persons without diabetes. Furthermore, the entry data showed that personality, the more stable or trait-dependent characteristic, was not related to glycemic control, whereas anxiety, depression, and quality of life variables, the dynamic or state-dependent characteristics, were significantly related to metabolic control. While not in the diagnostic pathologic range for these meas-

_B

.1 4 8 2 Q. ai

0

Q

-2

.= -

4

-10 I I I IH I I I I I I I I

i i I I i I I I I I I I I

- 4 - 2 0 2 4 6

- 4 - 2 0 2 4 6

Change in Delta HbA,( 0 /

364

o 40 & 30 £ 20 , ^ 10 S 0 °-IO §-20 -a-30 £-40 -50

i I I I I I I I I I I I I

- 4 - 2 0 2 4 6

FIG. 4. Correlation of changes in glycemic control and anxiety, depression, and quality of life. Changes in glycemic control were measured by the difference between the delta HbA, at entry and the 36th week. The result was expressed as a change in percentage. Changes in depression, anxiety, and PCL were measured by the difference in scale scores at entry and the 36th week.


PSYCHOLOGICAL AND SOCIAL CORRELATES OF GLYCEMIC CONTROL/R. S. MAZZE, D. LUC1DO, AND H. SHAMOON

TABLE 4 Randomized subjects on conventional and intensive treatment regimens Conventional 18th week HbA, delta Anxiety Depression

PCL 36th week HbA, delta Anxiety Depression

PCL

(N 3.7 1.5 3.6 6.9 (N 3.7 1.3 2.6 6.9

= ± ± ± ± = ± ± ± ±

18) 0.6* 0.5 0.9 2.1 15) 0.6 0.4 0.9 2.1

Intensive

(N 3.6 1.6 4.2 12.0 (N 4.0 1.9 3.0 10.1

= ± ± ± ± = ± ± ± ±

23) 0.3 0.4 1.0 4.0 19) 0.5 0.3 0.6 3.2

' ± Standard error.

ures, those in poorest glycemic control had significantly greater anxiety, depression, and problems of daily living than those in good control. Finally, as glycemic control changed with participation in the study, improvement in metabolic control was associated with improvement in anxiety, depression, and quality of life. Similarly, worsening metabolic control tended to be associated with increased anxiety, depression, and problems of daily living. We could associate none of these changes with the type of insulin therapy or glucose monitoring. Treatment per se on the conventional therapy (one to two injections with urine testing) versus on the intensive therapy (multiple injections with capillary blood testing) did not contribute differentially to metabolic control, anxiety, depression, or quality of life. Nor does it seem that selection of a therapy by random assignment versus by patient choice made a difference in terms of the outcome measures. Previous studies of persons with IDDM have either concentrated on the psychological responses to improved metabolic control with subjects using intensive therapies23 or assessment of psychological variables independent of therapeutic outcome.11 In the present study we examined a relatively homogeneous group of individuals (all persons with IDDM with no debilitating complications) who represented a broad spectrum of socioeconomic, racial, educational, and therapeutic backgrounds. To remove any differences that might arise from treatment, equal access to the health care team was provided to the subjects on both the conventional and intensive regimens with the overall goal of improving metabolic control. Thus, it is unlikely that these results are limited to only those patients following intensive regimens for glycemic control. Furthermore, since both the randomized and nonrandomized subject groups revealed the same trends, it is unlikely that the particular attributes of persons with diabetes willing to be randomly assigned to a treatment regimen accounted for these findings, or that patient preference for a particular form of treatment was an important factor. These data show that intensive therapeutic regimens per se do not produce adverse psychosocial results and that the degree of metabolic control may confer a beneficial effect on the psychosocial state of the person with diabetes.

The present study confirms an often held clinical view that the majority of persons with diabetes seem to be able to cope with their illness. Furthermore, our data do show that a significant linkage exists between dynamic or state-dependent psychosocial characteristics and metabolic control. These data support the notion that anxiety, depression, and quality of life are associated with glycemic control. At entry into our study, those in good, average, and poor control could be characterized differently according to dynamic psychosocial measures. Additionally, these characteristics were not intrinsic to the subject but rather were associated with metabolic control since substantial patient movement from one metabolic control category to another occurred throughout the study period. The clinical implication of these observations is that improved glycemic control carries with it improved quality of life, as well as decreased anxiety and depression. The current studies do not, however, permit us to conclude that a oneway relationship exists in which improved control of blood glucose necessarily leads to improvement in the quality of life of the person with diabetes. Indeed, the converse may be true. Improvement in the quality of life of the diabetic individual may foster improved metabolic control. Further studies are required if causality is to be established. Nevertheless, it should be noted that although we cannot presently establish the direction of this relationship, the study's clinical importance remains intact. ACKNOWLEDGMENTS: The authors wish to especially acknowledge the assistance of the members of the Diabetes Education Team of the DRTC, and Drs. Robert Plutchik and Hope Conte of the Department of Psychiatry, Drs. Howard Eder, Norman Fleischer, and Ronald Nagel of the Department of Medicine, and JoAnn Murphy of the DRTC. This study was supported by a pilot and feasibility study grant from the Einstein/Montefiore Diabetes Research and Training Center, National Institutes of Health grant #2P 60AM 20541-06 and a research and development contract from the Ames Division of Miles Laboratories.

From the Albert Einstein College of Medicine and Montefiore Medical Center, Diabetes Research and Training Center, 1300 Morris Park Avenue, Bronx, New York 10461. Address reprint requests to Dr. Roger Mazze at the above address. REFERENCES 1 Denolin, F., Appelbloom-Fondu, J., Lemiere, B., and Dorchy, H.: Psychological problems of diabetic adolescents: long-term follow-up. Pediatr. Adolesc. Endocrinol. 1982; 10:21-24. 2 Galatzer, A., and Laron, Z.: Psychological evaluation of newly diagnosed diabetics and their families. Pediatr. Adolesc. Endocrinol. 1982; 10:51-57. 3 Johnson, S. B.: Psychosocial factors in juvenile diabetes: a review. J. Behav. Med. 1980; 3:95-1174 Garmezy, N.: Behavioral issues in chronic illness. In Behavioral


365


and Psychosocial Issues in Diabetes. Hamburg, B., Lipsett, L., Inoff, G., and Drash, A., Eds. DHHS Pub. No. (NIH) 1980: 80-1193, 13-24. 5 Miller, J., Ill, Spitz, P., Simpson, U. ( and Williams, G. R: The social function of young adults who had arthritis in childhood. J. Pediatr. 1982; 100:378-82. 6 Teiramaa, E.: Psychosocial factors, personality and acute-insidious asthma. J. Psychosom. Res. 1981; 25:43-49. 7 Hauser, S. T., and Pollets, D.: Psychological aspects of diabetes mellitus: a critical review. Diabetes Care 1979; 2:227-32. 8 Slawson, F. P., Flynn, W. R., and Kollar, E. J.: Psychological factors associated with the onset of diabetes mellitus. JAMA 1963; 185:166-70. 9 Stein, S. P., and Charles, E. S.: Emotional factors in juvenile diabetes mellitus: a study of early life experience of adolescent diabetes. Am. J. Psychol. 1971; 128:700-704. 10 Grant, I., Kyle, G., Teichman, A., and Mendels, J.: Recent life events and diabetes in adults. Psychosom. Med. 1974; 36:12128. 11 Simonds, J. F.: Psychiatric status of diabetic youth matched with a control group. Diabetes 1977; 26:921-25. 12 Sullivan, B. J.: Self-esteem and depression in adolescent diabetic girls. Diabetes Care 1978; 1:18-22. 13 Jacobson, A. M., Hauser, S. P., and Noam, G.: Ego development in diabetes: a longitudinal study. Pediatr. Adolesc. Endocrinol. 1982; 10:1-8.

366

14 Skyler, J. S., Lasky, I. A., Skyler, D. L , Robertson, E. G., and Mintz, H. H.: Home blood glucose monitoring as an aid in diabetes management. Diabetes Care 1978; 1:150-57. 15 Sonksen, P. H., Judd, S. L , and Lowy, C : Home monitoring of blood glucose: method for improving diabetic control. Lancet 1978; 1:729-31. 16 Schiffrin, A., and Belmonte, M. M.: Comparison between continuous subcutaneous insulin infusion and multiple injections of insulin: a one-year prospective study. Diabetes 1982; 31:255-64. 17 Trivelli, L. A., Ranney, H. M., and Lai, N. T.: Hemoglobin components in patients with diabetes mellitus. N. Engl. J. Med. 1971; 284:353-57. 18 Plutchik, R., and Kellerman, H.: Emotions Profile Index Manual, Los Angeles, California, Western Psychological Service, 1975. 19 Plutchik, R., Williams, M., Jerrett, I., Karasu, T., and Kane, C : Emotions, personality and life stresses in asthma. J.Psychonomic Res. 1975; 22:425-31. 20 Zung, W.: A self-rating depression scale. Arch. Gen. Psychol. 1965; 12:63-67. 21 Mooney, R. L., and Gordon, L. V.: The Mooney Problem Checklist Manual. New York, The Psychological Corporation, 1950. 22 SAS System. Cary, North Carolina, SAS Institute Inc., 1983. 23 Jacobson, A., and Hauser, S.: Behavioral and psychological aspects of diabetes. In Diabetes Mellitus: Theory and Practice. Ellenberg, M., and Rifkin, H., Eds. New Hyde Park, N.Y., Medical Examination Publishing Co., 1982:1037-52.
