Social pathways to premenstrual symptoms

Research in Nursing 8 Health, 1995, 18, 225-237

Social Pathways to Premenstrual Symptoms Nancy Fugate Woods, Ellen S. Mitchell, and Martha J. Lentz

The purpose of this study was to explore feminine and menstrual socialization, expectations about experiencing symptoms, and the stressful nature of women’s lives among women with three perimenstrualsymptom patterns. Social learning and stress theory provided a theoretical framework for understanding why some menstruating women experience premenstrual syndrome or premenstrual magnification symptom patterns. Data about socialization, stressful life context, expectations about symptoms, depressed mood, and other health-related and demographic indicators were obtained from an interview. Subsequent daily recordings in a health diary for two or more menstrual cycles provided data with which to classify women’s symptoms across the menstrual cycle as a low severity symptom (LS, n = 73), premenstrual syndrome (PMS, n = 36), or premenstrual magnification (PMM, n = 62) pattern. Stepwise discriminant function analysis demonstrated that stressful life context, menstrual socialization, and expectations about symptoms related to menstruationdifferentiatedwomen with an LS from those with a PMS or PMM symptom pattern. In addition, depressed mood differentiated the three groups. 0 1995 John Wiley 8 Sons, Inc.

Despite a long tradition of inquiry about the social mediation of symptom experiences, most researchers have focused on the role of social stress in the generation of symptoms, placing relatively little emphasis on the role of social learning. The purpose of this study was to explore social pathways to a group of symptoms that women commonly experience and that are problematic for clinicians to diagnose and treat: premenstrual symptoms. Although a low severity (LS) symptom pattern across the menstrual cycle is the most prevalent symptom pattern in population-based research, a significant proportion of women experience the premenstrual syndrome (PMS) and premenstrual magnification (PMM) patterns (Mitchell, Woods, & Lentz, 1991; Woods, 1987). Women with the

PMS pattern have LS symptoms or no symptoms during the postmenses phase of their cycles and high severity symptoms during the premenses phase. Women with PMM, by contrast, have moderate to high severity symptoms during the postmenses phase that are exacerbated during premenses. Attempts to characterize premenstrual symptoms as a discrete disease entity with a clear biologic mechanism have proven futile in a series of investigations over the last 2 decades (see reviews by Rubinow & Schmidt, 1989, and Halbreich, Alt, & Paul, 1988). To date, only progesterone levels seem to be related to the course of symptoms, and only when comparisons are made within individuals with symptoms. Progesterone levels do not differ when groups of

Nancy Fugate Woods, PhD, RN, is a professor, School of Nursing, and Director, Center for Women’s Health Research, University of Washington. Ellen S. Mitchell, ARNP, PhD, and Martha J. Lentz, PhD, RN, are research associate professors, School of Nursing, University of Washington. This research was funded in part by grant NU 1054-03-06, National Center for Nursing Research, NIH. The authors acknowledge the assistance of Karen Allman, Lynne Buchanan, Cheri Cameron, Theresa Epstein, Barbara McGrath, and Barbara Polovitchwith data collection, and the participants for their generosity. This article was received on March 3, 1994, revised, and accepted for publication on October 17, 1994. Requests for reprints can be addressed to Dr. Nancy F. Woods, University of Washington, SM-23, Center for Women’s Health Research, Seattle, WA 98195. 0 1995 John Wiley & Sons, Inc. CCC 0160-6891/95/030225-13

225

226

RESEARCH IN NURSING & HEALTH

individuals with and without premenstrual symptoms are compared (Halbreich, Endicott, Goldstein, & Nee, 1986; Hammarback, Damier, & Backstrom, 1989; Lentz, et al., 1994; Rubinow et al., 1988). Indeed, inducing menses by administration of the antiprogestin agent mifepristone (which blocks actions of progesterone, produces regression of the corpus luteum, and terminates the luteal phase) has no effect on symptom experiences in women with PMS (Schmidt et a]., 1991). Recent attempts to understand the diffuse set of symptoms that some women experience around the time of menstruation increasingly reflect new models of this phenomenon. Conceptions of PMS as disease with a clearcut physiologic explanation have given way to thinking about the phenomenon as an instance of disease, and labels such as “premenstrual changes” have become commonplace in the scientific literature (Brooks-Gunn, 1986). Some have proposed that PMS can be understood as a state change in which the menstrual cycle choreographs a transition to a particular experiential state for some women. The state change is characterized by symptoms such as dysphoric mood and fluid retention as well as increased autonomic arousal and stress reactivity, changes in perceptions of and response to one’s environments, and perceived changes in behavior and cognitions. Instead of producing a specific symptom or group of symptoms, cyclic changes in ovarian steroids initiate a cascade of neuroendocrine events that alter stress perceptions and produce symptoms for some women (Hamilton, Alagna, & Sharpe, 1985; Rubinow & Schmidt, 1989). The state change hypothesis also accounts for women with premenstrual symptoms experiencing different arousal and stress reactivity patterns premenses and postmenses. Women with a PMS symptom pattern demonstrated a premenstrual increase in muscle tension and skin conductance levels at rest and in response to a cognitive stressor (Dickson-Parnell & Zeichner, 1989; Van den Akker & Steptoe, 1989), but women with a PMM pattern did not (Woods, Lentz, Mitchell, & Kogan, 1994). Women with premenstrual symptoms also respond to their environments differently during premenses than during other cycle phases. Women with menstrual-related mood disorders reported more negative life events than asymptomatic women (Schmidt, Grover, Hoban, & Rubinow, 1990). Moreover, they rated the same events as more distressing premenstrually than postmenses and rated fewer events as becoming more pleasant during premenses. Their

changes in ratings from postmenses to premenses were not related to changes in depression or anxiety (Schmidt et al., 1990). Neither women with an LS nor those with a PMS pattern rated cognitive stressors administered in a stress-testing laboratory (the Stroop test and solving anagrams) as more stressful premenses than postmenses (Woods, Lentz, & Mitchell, 1993, unpublished). Whether women with PMS and PMM symptom patterns experience changes in behavior or cognitions remains controversial. Although some women perceive performance or cognitive dysfunctions premenses, there is no evidence of actual performance or cognitive decrements for women with PMS (Rubinow & Schmidt, 1989). Indeed, Rodin ( 1976) demonstrated that women who were symptomatic around the time of menstruation actually performed more effectively than equally aroused nonmenstruating women and similarly to women who were not distressed at the time. Thus, women who experience predictable symptoms during premenses may actually bolster their performance of certain cognitive tasks. Although the state change model promises to be useful in understanding women’s premenstrual experiences, studies guided by the state change model have not yet accounted for the social context in which women experience menstruation. A common criticism of women’s health research by feminist theorists and scientists alleges that work not grounded in an understanding of the fabric of women’s lives is unlikely to be informative or serve emancipatory ends for women (Harding, 1991). Locating the problem of symptoms within the woman herself rather than acknowledging the simultaneous influences of women’s social experiences perpetuates the belief that biologic events that only women experience can be associated with premenstrual symptoms. A model that incorporates the social context in which menstruation occurs may more accurately reflect the realities of women’s lives and suggest possibilities for caring for women who find their symptoms distressing. A woman’s social context may contribute to and be affected by her experiences of affective and bodily changes, perceptions of and responses to her environments, and perceived personal and behavioral changes associated with her menstrual cycle. Dimensions of a woman’s social context include socialization about being a woman and, more specifically, her expectations about how she will experience menstruation. Another important dimension is the woman’s everyday stresses, strains, and resources with which to respond to them.

SOCIAL PATHWAYS TO PREMENSTRUAL SYMPTOMS / WOODS ET AL.

Social constructions about menstruation and premenstrual symptomatology influence women through social transmission of beliefs, attitudes, and expectations by health professionals, media, and women’s families and peers. Indeed, some contend that premenstrual syndrome reflects a social construction about menstrually related dysfunction (Rodin, 1992). Young girls learn about symptoms from observing their mothers, sisters, and peers, including expectations regarding menstrual experiences and effects of menstruation on feelings and behavior (Menke, 1983; Stoltzman, 1986). Learning stereotyped expectations about menstruation prior to menarche is common for young boys as well as girls (Clarke & Ruble, 1978). Moreover, young girls leam from the communications media that menstruation is a hygienic crisis to be concealed (Whisnant & Zegans, 197% and this model persists (Patterson & Hale, 1985). Mothers’ experiences with premenstrual symptoms appear to be linked to daughters’ subsequent symptom experiences and illness behavior (Freeman, Sondheimer, & Rickels, 1988; Menke, 1983; Stoltzman, 1986; Taylor, Woods, Lentz, Mitchell, & Lee, 1991; Whitehead, Busch, Heller, & Costa, 1986; Wilson, Turner, & Keye, 1991). Exposure to a mother with premenstrual symptoms, and teachings about negative effects of menstruation, were associated with negative affect symptomatology during the premenses and menses for adult women (Taylor et al., 1991). Moreover, women with PMS and PMM symptom patterns were more likely to have had a mother with more premenstrual symptomatology than women with an LS pattern (Mitchell, Woods, & Lentz, 1994). In a study of adult women with and without PMS, and their daughters, the number and severity of luteal phase symptoms were significantly greater for the women with PMS, and for their daughters, than for women without PMS, and for their daughters (Wilson et al., 1991). Women whose mothers had PMS and who were seeking treatment for PMS had more severe symptoms than those whose mothers did not (Freeman et al., 1988). Finally, mothers who exhibited sick role behavior during their menses and who encouraged their young daughters to adopt a sick role during their menses, such as staying home from school, had adult daughters who experienced more perimenstrual symptoms, made more clinic visits for symptoms, and had more days of restricted activity (Whitehead et al., 1986). Whether these findings reflect only social leaming or also genetic or physiologic similarity between mothers and daughters remains unclear.

227

Expectations about having symptoms are transmitted through socialization about menstruation. In a classic study, Ruble (1977) told college women that she could predict their next menses. Those who believed they were premenstrual, but were not, reported more symptoms than those who actually were premenstrual at the time of the study. Later studies showed that the effect of expectancy on premenstrual symptoms varied depending on the samples studied and the type of expectancy manipulated in the study. Young women’s expectancies that negative stereotypes accounted for PMS, as altered experimentally, lowered their subsequent reports of premenstrual negative mood, whereas expectancies about biologic causes had no effect on subsequently reported symptoms (Fradkin & Firestone, 1986). Stereotypic expectations of menstrual cycle symptoms influenced women’s symptom reports in a study in which women and men were exposed to variation in information about the purpose of the study and told that researchers had found specific symptoms occurred at different times of the menstrual cycle (Aubuchon & Calhoun, 1985). Although stereotypic biases in information processing about menstrual symptom cyclicity may influence symptom reporting (Koeske & Koeske, 1975; Ruble & Brooks-Gum, 1979), information processing about menstruation may be less important for women whose symptoms are severe. When women reporting severe symptoms meeting criteria for late luteal phase dysphoric disorder were studied, knowing that the purpose of the study involved PMS did not affect symptom ratings. Awareness of the study focus increased cyclicity in ratings of women with LS symptoms who were aware of the study when compared with those who were unaware. Nonetheless, the differences were small and most not statistically significant (Gallant, Popiel, Hoffman, Chakraborty, & Hamilton, 1992). In addition to learning about menstruation, socialization about women’s roles in the society provides a general cognitive orientation that may support more specific socialization about menstruation and its debilitating effects. Koeske and Koeske (1975) proposed that the simultaneous operation of a cognition linking negative moods and the premenstruum could result in women discounting situational factors and emphasizing biology in explaining their negative moods. This negative attributional style, in turn, could adversely affect women’s self esteem. Hamilton et al. (1985) pointed out that many women experience learned helplessness as an extension of life

228

RESEARCH IN NURSING B HEALTH

experiences related to their economic inequality, subordinate group status, vulnerability to victimization, situational stressors related to their roles, and the internalization of stereotypes that devalue women. Thus, women who have been exposed to difficult life circumstances that underscore how women are devalued may be more vulnerable to learning negative aspects about menstruation. Since the 1970s, researchers have examined relationships between naturally occurring stressors and perimenstrual symptoms. Early efforts demonstrated that major life stressors and daily hassles were related to perimenstrual symptoms, especially negative affect (Gannon, Luchetta, Pardie, & Rhodes, 1989; Siegel, Johnson, & Sarason, 1979; Taylor et al., 1991; Wilcoxon, Schrader, & Sherif, 1976; Woods, Dew, & Most, 1982; Woods, Most, & Longenecker, 1985). In more recent studies in which stringent criteria were used to select women with specific perimenstrual symptom patterns, daily stress ratings accounted for a modest amount of the variance in physical symptoms (6%) and in mood symptoms (10%) (Beck, Gevirtz, & Mortola, 1990). Women with severe symptoms meeting the criteria for PMS/late luteal phase dysphoric disorder did not differ from women without PMS, those using oral contraceptives, and men on ratings of frequency of stressful events. However, women with PMS rated their problems as more disturbing and demonstrated a cyclical pattern in negative interpersonal interactions at work (Gallant et al., 1992). Although most investigators have assumed that stressors caused or exacerbated perimenstrual symptomatology, women asserted that their symptoms precipitated stressful experiences such as interpersonal conflict and or increased sensitivity to their impact (Woods, Taylor, Lentz, & Mitchell, 1992). Of note is the absence of work exploring whether women see their social situations more clearly premenses than at other times of the cycle. The ability to perceive situations more sharply when aroused has been demonstrated as an adaptive component of stress response. Perhaps premenstrual arousal enhances women’s perceptions of situations such that their clearer perception premenses may be more accurate than their perceptions at other times. To date, little work has focused on how dimensions of women’s life context such as social support, employment, education, and income influence symptoms and stress during the premenstruum. Women who use social support as a coping strategy experienced less severe premenstrual symptoms (Warren & Baker, 1992).

Women with premenstrual symptoms or menstrually related mood disorders reported lower marital satisfaction than those without these symptoms, but the causal direction is unclear (Siegel, 1986; Stout & Steege, 1985; Winter, Ashton, & Moore, 1991). Employment does not seem related to premenstrual symptoms, but women in managerial jobs who experienced PMS perceived greater effects on their function than those in service jobs. Moreover, women managers received less social support from their superiors, co-wmkers, or spouses than did women in service jobs (Collins, 1991). The relationship between education and PMS is puzzling, with those with a PMS pattern having more formal education (Mitchell et al., 1994; Taylor et al., 1991). No relationship between income and PMS has been reported. Our earlier studies demonstrated that women with a PMS pattern had more pregnancies, experienced more depressed mood, perceived themselves as less healthy, performed more health practices, and were older than women with an LS symptom pattern. In addition, women with a PMM symptom pattern reported more severe depressed mood and were younger than those with either a PMS or LS symptom pattern (Mitchell et al., 1994). The purpose of the current analyses was to differentiate women with an LS, PMS, or PMM symptom pattern with respect to their feminine and menstrual socialization, expectations about menstrual symptoms, stressful life context, and resources for responding to stressors. These analyses are part of a larger investigation of the patterns of ovarian steroids, cortisol, and catecholamines, stress response, social demands, and social and personal resources among women with LS, PMS, and PMM symptom patterns (Woods et al., 1993).

METHOD

Sample Women were recruited through advertisements in local newspapers seeking both low-symptom and symptomatic women. A total of 1,149 women between the ages of 18 and 45 responded to newspaper advertisements. Women were excluded if they were using oral contraceptives or taking any other hormones, antidepressants, tranquilizers, diuretics, hypertension medications, or corticosteroids. Screening allowed only women who were not pregnant or lactating, who had at least one period in the past 3 months, who could un-

SOCIAL PATHWAYS TO PREMENSTRUAL SYMPTOMS I WOODS ET AL.

derstand and read English, who had no major gynecologic illnesses, and who reported over the past three cycles either minimal symptoms or symptoms that worsened premenstrually. A total of 522 women who met these criteria were interviewed in their homes and kept a daily health diary for 90 days (The Washington Women’s Health Diary) (Woods et al., 1993). Of these 522 women, 391 women returned a completed diary with at least two cycles of data. Each woman was assigned to a symptom pattern group based on a scoring system developed from a study of symptom patterns in a community-based sample of women (Mitchell et al., 1991). For the symptom pattern classification a mean severity score was obtained for five postmenses days (days 6 through 10) and for five premenses days (days -5 through -1) for two menstrual cycles using 33 symptoms from the Menstrual Symptom Severity List (MSSL) (Woods et al., 1993). Data presented here are from the 179 women who met the criteria for an LS, PMS, or PMM pattern based on two cycles of symptom ratings. Overall, women in the sample had a mean age of 34.8 years (SD = 6.4), a median annual income of $38,000, and a mean education of 15.4 years (SD = 1.7). Most were partnered and employed. Three women were Asian American, five were African American, five were Hispanic, and the remainder were Caucasian.

Measures Symptom patterns. Premenstrual symptom patterns were used as the indicator of the premenstrual state change. Based on their daily ratings of severity for 33 symptoms from the MSSL (Woods et al., 1993) for two cycles of data, women were classified as having an LS, PMS, or PMM pattern. The MSSL was developed from a large pool of symptoms commonly experienced during the perimenstruum. Subsequent analyses of symptom patterns experienced in a community-based sample of women and scoring using population-based criteria are described in detail elsewhere (Mitchell et al., 1991). The MSSL is contained within a longer list of 57 symptoms in the Washington Women’s Health Diary which are rated daily for occurrence and severity (Mitchell et al., 1991). Factor analysis of the multifactorial structure revealed consistent factors across menstrual cycle phases and internal consistency reliability estimates above .70 (Woods et al., 1993).

229

Stress and support need. Multiple indicators of the major constructs, feminine socialization, menstrual socialization, expectation of symptoms, stress, and support need, were included in the study. For purposes of data reduction, prior to comparing women with the LS, PMS, and PMM symptom patterns, data from these indicators, obtained from all women who participated in the initial interview (N = 488), were included in a principal components factor analysis with a varimax rotation. This procedure yielded five factors with eigenvalues exceeding 1.O. The measures used for each indicator are described subsequently followed by the factor analysis results. Socialization as a woman reflects the ways in which women were socialized about the roles of women in society, including in the workplace and their families. Feminine socializarion was assessed with the Attitudes Toward Women Scale (Spence & Helmreich, 1978) and Familism Scale (Paige, 1973). The Attitudes Toward Women Scale includes 15 items describing the roles, rights, and privileges women should have. Respondents used a 7-point scale in which responses ranged from agree strongly to disagree strongly. High scores on the scale indicated a profeminist, egalitarian attitude. Internal consistency estimates are high in this sample (alpha > .80). Familism, valuing the family’s wellbeing over the individual woman’s well-being, was assessed with the 7-item scale developed by Paige (1973). High scores indicate that a woman does not agree that her family’s well-being should take precedence over her own. Socialization for menstruation reflected what women had learned about menstruation through exposure to their mothers and sisters and through teaching about effects of menstruation. Menstrual socialization was indicated by the respondent’s rating of her mother’s perimenstrual symptoms, a sister’s symptoms, and the respondent’s recollection of her menarcheal preparation. Respondents were asked whether their mother or an adult woman in their home had any of these symptoms before or during menstruation: cramps, weight gain or swelling, headache, anxiety, tension, or irritability, backache, fatigue, painful breasts, food cravings, depression, crying, skin disorders, or mood swings. The total number of symptoms was used as an indicator of the mother’s symptoms. They were asked to rate the same list for their sister(s). Respondents also rated what they were told about effects of menstruation, as none, good, or bad effects. Items ( n = 12) were developed by Brooks-Gunn and Ruble (1980) and included getting along with friends or family,

RESEARCH IN NURSING & HEALTH

230

performance in sports, ability to concentrate or pay attention, going out to a party or on a date, taking part in class, how easily a girl gets tired, taking part in gym class, how easily a girl gets sick, how well a girl does on tests, how crabby or irritable a girl is, bicycle riding or swimming, and the kind or amount of food a girl eats. Bad effects were summed and used as an indicator of negative menstrual socialization in these analyses. Expectations about menstrual symptoms reflected how women expected to experience menstruation and symptoms. Indicators included seeing menstruation as debilitating, anticipation of menstruation, and self-rating of PMS. Current menstrual attitudes were measured with the Menstrual Attitudes Questionnaire Debilitation Scale, which taps women’s views of menstruation as a debilitating event with items such as becoming “more easily upset before and during a period than at other times” (Brooks-Gunn, Ruble, & Clark, 1977). Respondents rated each item on a 7-point Likert scale in which 1 indicated strongly disagree and 7 strongly agree. Cronbach’s alpha-

ratings are high (> .80) for this scale in this sample. Respondents also were asked two single items, one from the Menstrual Attitudes Scale tapping whether women could anticipate menses by their symptoms, the other whether they perceived they had PMS. Stress was measured using indicators of major life events and daily stressors. Life events were assessed with Norbeck’s ( 1994) woman-oriented revision of the Sarason Life Events Scale (LES). This 91-item scale contains events related to health, work, school, residence, love and marriage, family and close friends, personal and social, financial, crime, and legal matters and parenting. Participants indicated whether the event data were complete for the first interview (N = 488). Factor analysis results are presented in Table 1. Means and standard deviations for each variable used in the analysis for each symptom group are included in Table 2. Factor 1 (stress) included four variables: family-related stress, negative life events (LES scores), how stressed each woman felt over the

Table 1. Factor Loadings for indicators of Stress, Feminine Socialization, Support Need, Symptom expectation and Menstrual Socialization (N = 488) Factors Variables Stress Negative life events Stress, family Stress, amount Stress, personal Feminine socialization Attitudes, women Familism Support need Advice Affirmation Need relax Confidant Help Expectation of symptomsa Have PMS Anticipate symptoms Not debilitating Menstrual socialization Menstrual effects Mother’s symptoms Sister’s symptoms Eigenvalues % Variance

2

1

3

.70 .70 .67 .64

-.01 -.03 -.02 .04

.01

.90 .89

- .01

.20 .17 -.04 .38 -.06

-.lo

4

5

.21 .23

.07 -.19 - .09 - .01

.09 .04 - .08 .06

.04 .04

.03 .12

- .07 - .07

.09

-.lo

.09

.15

- .09

- .04

.14 -.lo - .34

.14 .09

-.04 .16 .17 -.03

.63 .62 .57 .54 .52

.23 .03 -.06

.04 .04 .34

- .09 - .06 - .06

- .73

.01 .04 -.01 2.85 16.8

.02 -.15 -.02 2.18 12.8

- .02

-

.04

-.15 - .02 1.35 7.9

-

.oo

aHigh scores = low symptom expectation.

.67 .60

-.13 1.26 7.4

-.18

.19 .04 - .29 .65 .64 .62 1.16 6.8

231

SOCIAL PATHWAYS TO PREMENSTRUAL SYMPTOMS I WOODS ET AL.

Table 2. Variables by Symptom Group (Means and SD)

FactorNariable Stress Negative life stress Family Stress How stressed Personal stress Feminine socialization Attitudes toward women Familism Need for support Affirmation Advice Confidant Relax Help Expectationa Anticipate symptoms Have PMS Menstruation debilitating Menstrual socialization Menstrual effects Mother’s symptoms Sister’s symptoms Other Pregnancies Age Incomeb Health perception Health-related stress Depressed mood Health practices Education=

Possible Score Range

LS

PMS

PMM

(n = 73) M (SD)

(n = 36) M (SD)

(n = 62) M (W

(0-237) (1-4) (0-3) (1-4)

8.0 1.9 1.4 2.0

(7.6) (1.0) (0.9) (0.9)

10.6 2.6 1.9 2.3

12.0 2.5 2.0 2.6

(15-105) (7-49)

87.3 47.4

(9.7) (7.3)

85.8 45.4

88.3 46.7

(1-3) (1-3) (1-3) (1-3) (1-3)

2.1 2.1 2.2 2.4 2.1

(0.6) (0.4) (0.5) (0.5) (0.5)

2.5 2.2 2.4 2.3 2.3

2.5 2.2 2.5 2.5 2.3

(1-7) (0-1) (7-77)

3.1 0.2 55.8

(0.9) (0.4) (7.3)

2.9 0.9 46.8

2.6 0.9 48.1

(0-24) (0-11) (0-11)

2.7 0.9 2.1

(2.3) (1.8) (2.8)

3.6 2.4 2.6

4.0 2.0 2.4

(0-12) (18-46) (1-27) (1-10) (1-4) (0-60) (1-8) (0-17)

1.4 36.7 6.3 8.9 1.4 5.6 5.1 5.9

(1.5) (5.3) (7.8) (1.1) (0.7) (5.3) (1.1) (1.4)

2.6 35.3 7.1 7.8 1.6 0.6 4.2 5.0

2.0 34.6 3.2 7.7 2.0 5.7 4.6 5.3

aHigh scores = low expectation. Wrdinal scale with 27 categories. COrdinal scale with 17 categories.

past 3 months, and her rating of personal stress. Factor 2 (feminine socialization) included ratings of familism and attitudes toward women. High scores on this factor reflected more contemporary attitudes toward women and their roles vis a vis their families. Factor 3 (support need) included women’s ratings of need for more affirmation, advice, access to a confidant, and someone to share fun and relax with them, and someone to help them. Factor 4 (symptom expectation) included three variables: anticipating symptoms, self-perceived PMS, and attitudes toward menstruation as debilitating. Low scores indicated high expectation of symptoms. Factor 5 (menstrual socialization) included exposure to a mother and/or sister with penmenstrual symptoms and having been taught negative effects of menstruation. Of all the indicators analyzed, none had

factor loadings of .40 or greater on any other factor. Each factor had an Eigenvalue exceeding I .o. Additional variables. Included in the factor structure derived as described earlier were a set of additional variables that had been included in our earlier study distinguishing women with premenstrual symptoms from those with a low severity symptom pattern. These included depressed mood, perception of health, health practices, number of confirmed pregnancies, and the personal resources of education, income, and age. The Center for Epidemiologic Studies Depression (CES-D) (Radloff, 1977) scale was used as an indicator of depressed mood. This 20-item scale requires respondents to rate how often certain symptoms occurred during the past week.

232

RESEARCH IN NURSING B HEALTH

Using a cutoff point of 16, the scale correctly identified 7 1% of clinically depressed patients (Newberry, Weissman, & Myers, 1979; Roberts & Vernon, 1983). The alpha for this study was .91. Health perception was measured by a selfrated assessment of one item on a 1 to 10 scale, where the highest score was a positive rating of one’s health. In addition, women indicated the number of pregnancies they had confirmed. Health practices were assessed with a modification of the Human Population Laboratory Questionnaire (Belloc & Breslow, 1972). Participants indicated their health behaviors (with positive behaviors in parentheses), including: cigarette smoking (no), consumption of alcohol-containing beverages (two or less per day), physical exercise (60to 120 minutes per week), body size (body mass index of 25 or less (computed as weight in kilograms/height in meters squared); and eating habits (three to four meals per day; one or fewer snacks per day) (Belloc & Breslow, 1972; Garrow & Webster, 1985; Wingard, Berkman, & Brand, 1982). A health practices score was calculated as the sum of each of the 8 items, with 1 representing the healthy behavior. Income was included in the analyses as an ordinal variable coded according to over 27 income categories from less than $5,000 to over $50,000. Each category represented $2,000. In addition, education was measured by years of formal education and age was recorded as current age in years.

Procedure Respondents were interviewed in their homes or place of their choosing as the initial aspect of this study and were paid for their participation. All data reported here were obtained during the faceto-face interviews after written informed consent was obtained. Data reduction was achieved by using principal components factor analysis with a varimax rotation as described earlier. Factors derived from the total sample (N = 488) were used to calculate factor scores, which were then entered along with indicators of health perception, health practices, depressed mood, pregnancies, income, education, and age into three 2-group discriminant function analyses to differentiate women with LS versus PMS, LS versus PMM, and PMS versus PMM patterns. Structure coefficients (discriminant loadings) represent the simple correlation of a variable with the discriminant function and are not affected by multicolinearity. The square of the structure coefficient indicates the proportion of the variance in the function attribut-

able to that variable. Canonical correlations, which indicate the amount of variance accounted for by the function, are provided as a basis for judging the substantive value of the function (Dillon & Goldstein, 1984).

RESULTS The function differentiating women with LS and PMS patterns included six variables with a structure coefficient near or exceeding .3 (Dillon & Goldstein, 1984) (Table 3). Expectation of symptoms, depressed mood, stress, confirmed pregnancies, health perception, and education differentiated the two groups. Women with PMS had greater expectation of having perimenstrual symptoms, higher depression scores, more stress in their lives, and more pregnancies, rated themselves as less healthy, and had less formal education than women with the LS pattern. Although the structure coefficient was less than .3, menstrual socialization also contributed significantly to this function, such that women with a PMS pattern had more negative menstrual socialization. This function explained 73% of the variance. Eighty-five percent of the women were correctly classified with similar prediction for the LS and PMS groups (LS = 85%; PMS = 86%). Expectation of symptoms, depressed mood, stress, health perception, and menstrual socialization differentiated the women with LS and PMM patterns. Those with PMM demonstrated greater expectation of perimenstrual symptoms, higher depression scores, more negative health ratings, and higher stress levels, and were exposed to more negative socialization about menstruation (Table 3) than the women with an LS pattern. This function explained 72% of the variance and correctly classified 85% of the women (PMS = 73%; PMM = 74%). Depressed mood, expectation of symptoms, stress, and menstrual socialization differentiated the women with a PMS and PMM pattern. Women with a PMS pattern had lower scores on the depression scale, fewer expectations of symptoms, less stress, less negative menstrual socialization, and higher incomes than women with the PMM pattern. The function explained only 53% of the variance and correctly classified 73% of the women, with a similar classification of the women with a PMS pattern (73% vs. 74%). These analyses provide support for inclusion of social as well as behavioral and biologic factors in a model distinguishing women who experience premenstrual symptoms (those with a PMS

=

- 1.49

.74

.73

.29 .28' - .24' .17 - .09' .07 .06* - .01

- .33* - .32'

- .36

.68'

Structure Coefficient

109)

=

.w

.70'

.96 -1.10

.72

.31 - .27* .19' .16' .13 -.11* .10 - .05* .01

- .36*

-

Structure Coefficient

premenstrual magnification.

Canonical correlation Centroids LS PMM

Symptom expectation (4) Depression Stress (1) Health perception Menstrual socialization Age Income Education Confirmed pregnancy Health practices Support need (3) Feminine socialization (2)

FactorNariable

LS VS. PMM (n = 133)

Note. LS = low symptom; PMS = premenstrual syndrome; PMM aNumber in parentheses refers to factor number. ' p < .05.

Canonical correlation Centroids LS PMS

Symptom expectation (4)a Depression Stress (1) Confirmed pregnancy Health perception Education Menstrual socialization (5) Health practices Support need (3) Income Feminine socialization (2) Age

FactorNariable

LS VS. PMS (n

Canonical correlation Centroids PMS PMM

Depression Symptom expectation (4) Stress (1) Menstrual socialization (5) Income Health perception Health practices Feminine socialization (2) Age Support need (3) Education Confirmed pregnancy

FactorNariable

.81

- .47

.53

.oo

.40' .31* - .29* - .27 -.24 .18' -.18 .11 -.11

- .62*

.76'

Structure Coefficient

PMS VS. PMM (n = 168)

Table 3. Discriminant Analyses DifferentiatingWomen with LS, PMS, and PMM Symptom Patterns

RESEARCH IN NURSING 8 HEALTH

234

or PMM pattern) from those who do not (LS pattern). The ability to distinguish women with PMS and PMM patterns using the proposed model is limited. This may be a function of limited power, given the relatively modest sample sizes in these two groups, or it may reflect common social pathways for both symptom patterns. DISCUSSION Expectation of symptoms distinguished women with both PMS patterns from those who had LS symptoms premenses, and those with a PMS from those with a PMM pattern. Expectation of symptoms linked to the menstrual cycle has been associated with physiologic arousal premenses, another dimension of the state change model (Hamilton et al., 1985; Rubinow & Schmidt, 1989). Coyne (1983) and Van den Akker and Steptoe (1989) found that women who anticipated, but did not necessarily experience, the most severe premenstrual symptoms demonstrated elevated EMG levels premenses. Women who experience recurrent premenstrual symptoms may accurately anticipate increased premenses arousal and symptoms. Of interest is that women with the LS symptom pattern had the lowest expectation of symptoms, followed by those with the PMM and PMS patterns, respectively. In the stress laboratory assessments conducted postmenses and premenses with a subset of our same sample, those with an LS pattern exhibited the lowest levels of physiologic arousal premenses, followed by those with a PMM and PMS symptom pattern, respectively (Woods et al., 1994). Thus, an increasing level of expectation was associated with increased evidence of physiologic arousal as well as with a symptom pattern. It is possible that as women experience symptoms premenses they learn to anticipate them and a cyclic pattern evolves that continues to link expectation to arousal and symptoms. A stressful life context also distinguished women with a LS, PMS, and PMM. Those with PMS or PMM reported more negative life event stress over the past year than LS women, as well as more recent stress related to family matters and personal issues, consistent with our earlier findings in a population-based sample (Taylor et al., 1991; Woods et al., 1985). Sensitization to repeated stressors can produce dysphoric mood and can play a role in recurrent affective disorders (Gold, Goodwin, & Chrousos, 1988; Post & Ballenger, 1981). Thus, cyclic symptoms may become linked to expectations of symptoms and

to a generally stressful life context. Post hoc analyses revealed that women who experienced greater stress were more likely to expect to experience premenstrual symptoms ( r = .27, p < .05). Socialization for menstruation through exposure to mothers and sisters who had perimenstrual symptoms and exposure to negative teachings about menstruation also differentiated women with the three different symptom patterns. This distinction is consistent with a social learning theory explanation linking social and cultural exposures to subsequent experiences of symptoms and illness behavior, and is consistent with our earlier findings (Taylor et al., 1991; Woods et al., 1992) and those of others (Freeman et al., 1988; Menke, 1983; Stoltzman, 1986; Whitehead et al., 1986; Wilson et al., 1991). Whether socialization as well as genetic and physiologic resemblance between mothers and daughters act jointly to influence symptoms remains to be determined. Health promotion interventions, including health education about the consequences of menstruation for young women, may benefit from orienting their messages about menstruation in a more positive fashion. Mothers’ attributions of their distress to the menstrual cycle may also encourage young women to locate the source of their own distress within their bodies rather than exploring social conditions that contribute to their distress. In our earlier work, exposure to a mother with PMS symptoms was also linked to women’s current experience of depressed mood, suggesting that socialization experiences may contribute to young women’s acquiring a propensity for depressed mood (Taylor et al., 1991). Depressed mood scores also distinguished the three groups of women. As anticipated, those with the LS pattern had the lowest depression scores. Women with the PMM pattern had the highest CES-D scores, with their mean score approaching 16, the score differentiating those with clinical depression from controls in earlier validation studies (Newberry et al., 1979; Roberts & Vernon, 1983). Depression has been associated with PMS in other studies in which women with a history of major depression demonstrate a higher incidence of PMS (Clare, 1983; Halbreich & Endicott, 1985). In our sample women with a PMS pattern and those with an LS pattern had CES-D scores resembling those for healthy samples (Radloff, 1977). Future work is needed to determine whether women who have a history of depression are more likely to experience the PMM symptom pattern rather than the PMS pattern.

SOCIAL PATHWAYS TO PREMENSTRUAL SYMPTOMS / WOODS ET AL.

The higher prevalence of depressed mood among women with PMS reported elsewhere may be attributable to misclassification of women with a PMM pattern as having a PMS pattern. In addition, common pathways to depression and PMS or PMM patterns may account for our findings. These may include social pathways such as social learning or biologic pathways as reflected in individual differences in response to changing ovarian steroids across the menstrual cycle and over the lifespan. Pregnancy also differentiated women with a PMS and LS pattern but was not important in discriminating women with an LS versus PMM or PMS versus PMM pattern. This finding is consistent with earlier work linking number of pregnancies and number of children of PMS (Freeman et al., 1988; Woods, Most, & Dery, 1982). Whether the pregnancy effect is a biologic one linking women’s previous endocrine events to mood change, or a social one operating through effects of parenting, remains to be determined. Post hoc analyses support the latter interpretation: number of confirmed pregnancies tended to be correlated with stress ( r = .29, p < .07). Health perceptions differentiated women with LS from those with a PMS or PMM pattern. Of importance is that all three groups saw themselves as quite healthy, consistent with judgments of women participating in our earlier studies (Woods et al., 1992). Taken together, these results provide support for a socially mediated model of premenstrual symptoms that includes the influence of expectations about symptoms, stressful life circumstances, menstrual socialization, and depressed mood. Women’s experiences of a stressful life context and socialization about menstruation may influence their expectations about symptoms, and their experiences of symptoms may in turn reinforce their expectations about menstruation as a symptomatic experience. Indeed, social mediation of symptomatology, through social learning and the stressful nature of a respondent’s life context, may play an important role in symptoms that are linked to biologic changes occumng with the menstrual cycle. These findings support development of interventions for young women that offer alternative attributions for their distress, as well as stress management strategies designed to reduce the distress some women experience premenses. Women with the PMM symptom pattern may benefit from therapies directed at their depression, as well as the strategies more specific to premenstrual symptoms. More careful examination of the context for women’s lives may provide

235

clues to reducing the prevalence of depressed mood as well as premenstrual symptoms.

REFERENCES Aubuchon, P.G., & Calhoun, K.S. (1985). Menstrual cycle symptomatology: The role of social expectancy and experimental demand characteristics. Psychosomatic Medicine, 47, 35-45. Barerra, M. (1981). Social support in the adjustment of pregnant adolescents: Assessment issues. In B. Gottlieb (Ed.), Social networks and social support (pp. 69-96). Beverly Hills: Sage. Beck, L., Gervitz, R., & Mortola, J. (1990). The predictive role of psychosocial stress on symptom severity in premenstrual syndrome. Psychosomatic Medicine, 52, 536-543. Belloc, J., & Breslow, L. (1972). Relationship of physical health status and health practices. Preventive Medicine, I , 409-42 1. Brooks-Gunn, J. (1986). Differentiating premenstrual symptoms and syndromes. Psychosomatic Medicine, 48, 385-387. Brooks-Gunn, J., & Ruble, D. (1980). The menstrual attitudes questionnaire. Psychosomatic Medicine, 42, 503-512. Brooks-Gunn, J., Ruble, D., & Clarke, A. (1977). College women’s attitudes and expectations concerning menstrual related changes. Psychosomatic Medicine, 39, 288-298. Clare, A. (1983). Psychiatric and social aspects of premenstrual complaint. Psychological Medicine Monograph-Supplement, 4 , 1-5. Clarke, A., & Ruble, D. (1978). Young adolescents’ beliefs concerning menstruation. Child Development, 49, 231-234. Collins, A. (1991). Premenstrual distress: Implications for women’s working capacity and quality of life. In M. Frankenhauser, U. Lundberg, & M. Chesney ( U s . ) , Women, work, and health: Stress and opportunities (pp. 239-254). New York: Plenum. Coyne, C. (1983). Muscle tension and its relation to symptom in the premenstruum. Research in Nursing & Health, 6 , 199-205. Dickson-Parnell, B., & Zeichner, A. (1989). The premenstrual syndrome: Psychophysiologic concomitants of perceived stress and low back pain. Pain, 34, 161-169. Dillon, W., & Goldstein, M. (1984). Multivariate analysis: Methods and applications. New York: Wiley. Fradkin, B., & Firestone, P. (1986). Premenstrual tension, expectancy, and mother-child relations. Journal of Behavioral Medicine, 9 , 245-249. Freeman, E. W., Sondheimer, S.J., & Rickels, K. (1988). Effects of medical history factors on s y m p tom severity in women meeting criteria for premenstrual syndrome. Obstetrics & Gynecology, 72, 236-239.

Gallant, S . J., Popiel, D.A., Hoffman, P.K., Chakraborty, P.K., & Hamilton, J.A. (1992). Using daily

236

RESEARCH IN NURSING 8 HEALTH

ratings to confirm premenstrual syndrome/late luteal phase dysphoric disorder. Part 1. Effects of demand characteristics and expectations. Psychosomatic Medicine, 54, 149-166. Cannon, L., Luchetta, T.,Pardie, L., & Rhodes, K. (1989). Perimenstrual symptoms: Relationships with chronic stress and selected lifestyle variables. Behavioral Medicine, 12, 149-159. Garrow, J., & Webster, J. (1985). Quetelet’s index as a measure of fatness. Internal Journal of Obesity, 9 , 147- 153. Gold, P., Goodwin, F., & Chrousos, G. (1988). Clinical and biochemical manifestations of depression: Relation to the neurobiology of stress. New England Journal of Medicine, 319, 413-420. Halbreich, U . , Alt, I.H., & Paul, L. (1988). Premenstrual changes: Impaired hormonal homeostasis. Neurological Clinics, 6, 173- 194. Halbreich, U., & Endicott, J. (1985). Methodological issues in studies of premenstrual changes. Psychoneuroendocrinology, 10, 15-32. Halbreich, U . , Endicott, J., Goldstein, S., & Nee, J. (1986). Premenstrual changes and changes in gonadal hormones. Acta Psychiatrica Scandinavica, 74, 576-586. Hamilton, J., Alagna, S . , & Sharpe, K. (1985). Cognitive approaches to understanding and treating premenstrual depression. In H. Ofofsky (Ed.), Premenstrual syndrome (pp. 69-84). Washington, DC: American Psychiatric Press. Hammerback, S., Damier, J.E., & Backstrom, T. (1989). Relationship between symptom severity and hormone changes in women with premenstrual syndrome. Journal of Clinical Endocrinology Metabolism, 68, 125-130. Harding, S. (1991). Whose science? Whose knowledge? Thinking from women’s lives. Ithaca, NY: Cornell University Press. Koeske, R.K., & Koeske, G.F. (1975). An attributiona1 approach to moods and the menstrual cycle. Journal of Personality and Social Psychology, 31, 474478. Lentz, M.J., Henker, R., Heitkemper, M., Mitchell, E.S., Woods, N.F., & Shaver, J. (1994). The relationship between ovarian steroids and perimenstrual symptoms: A comparison of group diferences and intra-individual patterns. Manuscript submitted for publication. Menke, E. (1983). Menstrual beliefs and experiences of mother-daughter dyads. In S. Golub (Ed.), Menarche (pp. 133-137). New York: Lexington. Mitchell, E.S., Woods, N.F., & Lentz, M.J. (1991). Recognizing PMS when you see it: Criteria for PMS sample selection. In D.L. Taylor & N.F. Woods (Eds.), Menstruation, health, and illness (pp. 89118). Washington, DC: Hemisphere. Mitchell, E., Woods, N., & Lentz, M. (1994). Differentiation of women with three perimenstrual symptom patterns. Nursing Research, 43, 25-30. Newbeny, P., Weissman, M., & Myers, J. (1979). Working wives and housewives: Do they differ in

mental states and social adjustment? American Journal of Orthopsychiatry, 49, 282-291. Norbeck, J. (1984). Modification of life event questionnaires for use with female respondents. Research in Nursing & Health, 7 , 61-71. Paige, K. (1973). Women learn to sing the menstrual blues. Psychology Today, 7 , 41-46. Patterson, E., & Hale, E. (1985). Making sure: Integrating menstrual care practices into activities of daily living. Advances in Nursing Science, 7 , 18-31. Post, R., & Ballenger, J. (1981). Kindling models for the progressive development of behavioral psychopathology: Sensitization to electrical, pharmacological and psychological stimuli. In Van Praag, H., Lader, M., & Rafaelsen, 0. (Eds.), Handbook of biological psychiatry (pp. 609-65 1). New York: Marcel-Dekker. Radloff, L. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, I , 385401. Roberts, R., & Vernon, S . (1983). The Center for Epidemiological Studies Depression Scale: Its use in a community sample. American Journal of Psychiatry, 140, 41-107. Rodin, J. (1976). Menstruation, reattribution, and competence. Journal of Personality, 33, 345-353. Rodin, M. (1992). The social construction of premenstrual syndrome. Social Science and Medicine, 13, 49-56. Rubinow, D.R., Hoban, M.C., Grover, G.N., Galloway, D.S., Roy-Bryne, p., Andersen, R., & Merriam, G.R. (1988). Changes in plasma hormones across the menstrual cycle in patients with menstrually related mood disorder in control respondents. American Journal of Obstetrics-Gynecology, 158, 5-11. Rubinow, D.R., & Schmidt, P.J. (1989). Models for the development and expression of symptoms in premenstrual syndrome. Psychiatric Clinics of North America, 12, 53-68. Ruble, D. (1977). Premenstrual symptoms: A reinterpretation. Science, 197, 291-292. Ruble, D.N., & Brooks-Gunn, J. (1979). Menstrual symptoms: A social cognition analysis. Journal of Behavioral Medicine, 2 , 179-195. Schmidt, P.J., Nieman, L.K., Grover, G.N., Muller, K.L., Memam, G.R., & Rubinow, D.R. (1991). Lack of effect of induced menses on symptoms in women with premenstrual syndrome. New England Journal of Medicine, 324, 1174-1 179. Schmidt, P., Grover, G., Hoban, C., & Rubinow, D. (1990). State-dependent alterations in the perception of life events in menstrual-related mood disorders. American Journal of Psychiatry, 147. 230-234. Siegel, J. (1986). Marital dynamics of women with premenstrual tension syndrome. Family Systems Medicine, 4 , 358-367. Siegel, J., Johnson, J., & Sarason, I. (1979). Life changes and menstrual discomfort. Journal of Human Stress, 5 , 41-46.

SOCIAL PATHWAYS TO PREMENSTRUAL SYMPTOMS / WOODS ET AL.

Spence, J., & Helmreich, R. (1978). Masculinity and femininity: Their psychological dimensions, correlates, and untecedents. Austin: University of Texas Press. Stoltzman, S. (1986). Menstrual attitudes, beliefs, and symptom experiences of adolescent females, their peers, and their mothers. Health Care for Women International, 7. 97- 1 14. Stout, A.L., & Steege, J.F. (1985). Psychological assessment of women seeking treatment for premenstrual syndrome. Journal of Psychosomatic Research, 29, 621-629. Taylor, D., Woods, N.F., Lentz, M., Mitchell, E.S., & Lee, K. (1991). Penmenstrual negative affect: Development and testing of an explanatory model. In D. Taylor & N. Woods (Eds.), Menstruation: Health and illness (pp. 103-1 18). Washington, DC: Hemisphere. Van den Akker, O., & Steptoe, A. (1989). Rychopysiological responses in women reporting severe premenstrual symptoms. Psychosomatic Medicine, 51, 319-328. Warren, C.J., & Baker, S. (1992). Coping resources of women with premenstrual syndrome. Archives of Psychiatric Nursing, 6, 48-53. Whisnant, L., & Zegans, L. (1975). A study of attitudes toward menarche in white middle-class American adolescent girls. American Journal of Psychiatry, 132, 809-814. Whitehead, W.E., Busch, C.M., Heller, B.R., & Costa, Jr. (1986). Social learning influences on menstrual symptoms and illness behavior. Health Psychology, 5 , 13-23. Wilcoxon, L., Schrader, S., & Sherif, C. (1976). Daily self reports on activities, life events, moods,

237

and somatic changes during the menstrual cycle. Psychosomatic Medicine. 38, 399-4 17. Wilson, C.A., 'hmer, C.W., & Keye, W.R. (1991). Firstborn syndrome: A comparison. Journal of Adolescent Health, 12, 130-137. Wingard, D.L., Berkman, L.F., & Brand, R.J. (1982). A multivariate analysis of health-related practices. American Journal of Epidemiology, 116, 765-775. Winter, E.J.S., Ashton, D.J., & Moore, D.L. (1991). Dispelling myths: A study of PMS and relationship satisfaction. Nurse Practitioner, 5 , 37-45. Woods, N. (1987). Penmenstrual symptoms: Another look. Public Health Reports, Supplement (JulyAugust), 102, 106-1 12. Woods, N., Dery, G., & most, A. (1982). Stressful life events and perimenstrual symptoms. Journal of Hum a n Stress, 8 , 23-31. Woods, N., Lentz, M., & Mitchell, E. (1993). Prevalence of perimenstrual symptoms. Unpublished report to National Center for Nursing Research. Woods, N., Lentz, M., Mitchell, E., & Kogan, H. (1994). Arousal and stress response across the menstrual cycle in women with three perimenstrual symptom patterns. Research in Nursing Q Health, 17, 99-110. Woods, N., Most, A., & Dery, G. (1982). Prevalence of penmenstrual symptoms. American Journal of Public Health, 72, 1257-1264. Woods, N., Most, A., & Longenecker, G. (1985). Major life events, daily stressors, and perimenstrual symptoms. Nursing Research, 34, 263-267. Woods, N., Taylor, D., Lentz, M., & Mitchell, E. (1992). Penmenstrual symptoms and health seeking behavior. Western Journal of Nursing Research, 14, 418 - 4 3 ,