Predicting psychological morbidity in Chinese women after surgery for breast carcinoma

Abstract

BACKGROUND

Between 30% and 70% of western women experience psychological morbidity after undergoing surgery for breast carcinoma; however, the rates and risk factors among Chinese women are unknown. Identifying at-risk women enables preventive intervention.

METHODS

Among 430 Chinese women who were approached within 1 week after undergoing surgery for early-stage breast carcinoma (baseline), 405 women (94%) completed measures of self-efficacy and psychological morbidity (the Chinese Health Questionnaire 12-item instrument [CHQ12]) and completed retrospective measures of treatment decision-making (TDM) difficulties, satisfaction with TDM involvement, and satisfaction with consultation and treatment outcome expectations. One-month postsurgery follow-up (follow-up), CHQ12 scores for 367 of 405 women (91%) were adjusted for concurrent physical symptom distress and trait optimism (the revised Chinese Life Orientation Test) and baseline predictors using stepwise multivariate regression.

RESULTS

At baseline 28% of women evidenced mild psychological morbidity, and 42% of women evidenced moderate-to-severe psychological morbidity: At follow-up, the respective rates were 32% and 36%. Preferred TDM involvement was associated with lower psychological morbidity (F = 6.702; P < 0.001). Baseline CHQ12 scores were predicted by outcome expectancies and TDM difficulties (adjusted regression coefficient [R²] = 0.192). Baseline CHQ12 scores and follow-up chemotherapy, in turn, predicted physical symptom distress at follow-up. After adjustment, high physical symptom distress, baseline psychological morbidity, low optimism, and no chemotherapy independently predicted follow-up CHQ12 scores (adjusted R² = 0.585).

CONCLUSIONS

Psychological morbidity was linked to women's TDM difficulties, their inability to anticipate treatment effects accurately, and physical symptom distress, possibly exacerbated by symptom misattribution. Optimizing TDM support and helping women accurately determine outcomes in terms of symptom experience and meaning and physical appearance may help to reduce psychological morbidity. Women who have TDM difficulties should be considered to be at high risk for psychological distress. Cancer 2005. © 2004 American Cancer Society.

Breast carcinoma (BC) and its treatments precipitate significant psychological distress. After BC surgery, depression rates ranging from 9% to 55% have been reported in previous studies, which primarily focused on Caucasian women.1–5 “Depression” among Mainland Chinese women exceeds 25%.6 Psychological morbidity in women with BC peaks 1–3 months postsurgery7 and persists for > 16 months. Social disruption affects 10% of women, and reduced attractiveness affects 22% of women, for at least 5 years.8 Some of this variation in rates is due to methodological and diagnostic differences, but specific variables also have been associated with greater psychological morbidity. Body-image disruption has been implicated primarily, with morbidity reportedly highest after mastectomy (MRM) and lowest after breast-conserving treatment (BCT).9–11 Symptom distress and impaired functioning from surgery, adjuvant chemotherapy, and radiotherapy also contribute to psychological morbidity.5, 8, 12 It is unclear whether women with preexisting depression report more symptom distress. Finally, coping,13–15 optimism,13–16 and self-efficacy for coping17, 18 all mediate cancer distress in western studies. Additional predictors and moderators, particularly culturally specific factors, also may be culpable. Potential factors include consultation; decision making; uncertainty and treatment practices; patient values, expectations, and beliefs about cancer; and family and community norms and responses.

In BC, research studies on western women predominate. Psychological morbidity in Chinese women with BC has not been explored, although Chinese, the largest ethnic grouping, comprise significant communities worldwide. Chinese societies are collectivist, with the individual's needs subordinated to the family or group. Conformity and “normality” support social harmony, which is a dominant social motive and a core value. Hong Kong (HK) Chinese women (hereinafter referred to as Chinese women) with BC differ from western women. Excepting survival, they express different outcome priorities, including social harmony and role functions.19 BCT is chosen more commonly among western women than among Chinese women.20 Although younger western women more often chose BCT, younger Chinese women often prefer MRM plus reconstruction (MRM+R),21 possibly because, in smaller breasted women, BCT may produce inferior cosmetic results.19, 22 However, Chinese women also emphasize rapid renormalization, and they perceive MRM as way to avoid further treatment.22 Fear of rejection underpins Chinese women's body-image concerns.19, 22 Most Chinese women who face BC express preferences for shared treatment decision-making (TDM), although, usually, they must chose for themselves their surgical treatment under conditions of high uncertainty.11, 19, 21, 22

Therefore, we hypothesized that more psychological morbidity (distress), both in the immediate postoperative period and 1 month later, would be reported by Chinese women with BC who 1) hold more negative outcome expectations and 2) report more TDM difficulties, reflecting higher decisional conflict. In addition, 3) because TDM difficulties are a function of satisfaction with TDM involvement,21 optimal TDM involvement should predict lower baseline psychological distress (Hypothesis 1). After the receipt of pathology reports on tumor stage and lymph node status between 2 weeks and 4 weeks postoperatively, the onset of adjuvant treatment (chemotherapy, radiotherapy), and facing adjustment to body-image disruption and altered form, we hypothesized that, at 1 month postsurgery, women's psychological distress would be related to their negative outcome expectations, treatment factors, optimism, and self-efficacy (Hypothesis 2). If they were significant, then these predictors of distress should remain after adjustment for demographic, operative, and medical treatment contributions (Hypothesis 3), support-group treatment, and the impact of the pathology report (Hypothesis 4).

MATERIALS AND METHODS

HK has a high per capita income, a pluralistic medical system, and disease patterns similar to those of other postindustrial countries. BC incidence was 37.9 per 100,000 women in 1997–1999,23 with 1918 new diagnoses reported in 2000.24 Of the 6.7 million population, mostly immigrants from the southern and eastern China coastal regions, Shanghainese, Vietnamese Chinese, and Hakka, among others, 98% are Chinese, and through many diasporas, most countries have significant Cantonese-speaking communities.

Patients, Setting, and Procedure

We initially identified key concerns postsurgery among HK Chinese women with BC.13, 16 Then, the current study was initiated to examine the hypotheses described above. After institutional ethics approval was obtained, all women who underwent surgery for Stage O-IV BC in 6 regional HK public hospitals between October, 2001, and January, 2003, were approached within 5 days after surgery (baseline) and were invited to participate. Participants immediately completed baseline assessments and were followed 1 month later. Inclusion criteria were age > 19 years and Cantonese fluency. Exclusion criteria were linguistic or intellectual difficulties, an active Axis I psychiatric diagnosis, or uncontrolled metastatic brain disease. Among 480 women, 447 women (93%) were available and eligible, and 405 women (91%) provided fully informed consent and completed a baseline face-to-face interview assessment. At 1 month postsurgery, these women completed a telephone interview follow-up assessment (follow-up).

Assessment

Psychological morbidity was assessed by the Chinese Health Questionnaire (CHQ12), which is a 12-item, Taiwanese validated version of the Goldberg General Health Questionnaire. Developed as a population screen for mild psychological morbidity in Chinese, its Likert scale is scored as follows: 0 = “not at all,” 1 = “about the same as usual,” 2 = “more than usual,” and 3 = “much more than usual.”25 Possible scores for the CHQ12 range from 0 to 36, sensitivity is 78%, specificity is 77%,26 and the Cronbach α = 0.84.27 Patient criterion is met when scores exceed 4.25 Scored between 5 and 10 indicate mild morbidity, and scores > 10 indicate moderate/severe morbidity. The CHQ12 was chosen, because we expected mild-to-moderate distress, and the anticipated “floor” effects eliminated more sensitive instruments. In addition, at follow-up, women were in community settings.

Optimism (i.e., the tendency to hold positive outcome expectancies) and self-efficacy (i.e., the confidence in one's ability to achieve specific goals and to deal with difficulties; important in predicting coping effort) indicate coping resilience. Dispositional optimism was assessed using the revised Chinese Life Orientation Test (C-LOT-R).28 Comprised of six-items scored on a four-point Likert scale from “strongly disagree” to “strongly agree,” higher scores reflect greater trait optimism. Dispositions theoretically are stable and can be measured reliably at any time. The Chinese Generalized Self-efficacy Scale (GSeS) was used to measure general self-efficacy.29 The 10 GSeS items are personal response-oriented (e.g., “I can solve most problems if I invest the necessary effort,”) and are scored on a 4-point Likert scale from “exactly true” to “not at all.” Higher scores reflect greater self-efficacy. Both instruments were validated in the HK population.28, 29 Among HK Chinese women with BC, factor analyses revealed stable factors.29, 30

TDM difficulties reflect patient uncertainty about treatment outcomes. Women who are more confident that they can predict treatment consequences accurately should show fewer TDM difficulties and less distress. Therefore, we measured aspects of TDM and expected treatment outcome. Women were asked at baseline to rate expected impacts of treatment outcomes (outcome expectancies) using the modified Breast Cancer Treatment Decision-Making Questionnaire (BCTDMQ). The 13 items of the BCTDMQ address expectations of negative surgical impact on respondents' sense of femininity, appearance, sexuality, normality, attractiveness, routine, likelihood of recurrence, additional treatment and reconstruction, speed of recovery, partner support, and cure.30, 31 Scored on a five-point Likert scale from “none at all” to “very much,”, higher BCTDMQ scores reflect greater expected impact.

To assess decision-making difficulty (TDM difficulties), women completed eight four-point Likert items scored from “strongly disagree” to “strongly agree,” six items measuring perceived TDM barriers (e.g., “shocked—could not think what to do,” “could not recall information,” and “did not know what questions to ask”), and two TDM facilitator items (adequacy of time and information). Higher scores reflected greater TDM difficulties.30 Women indicated the perceived degree of TDM involvement as follows: 1) “I was not allowed to participate,” 2) “less participation than I wanted,” 3) “as much participation as I wanted,” 4) “more participation than I wanted,” and 5) “much more participation than I wanted” (TDM involvement). Finally, women rated satisfaction with physician communications using an adapted Medical Information Satisfaction Scale (MISS).32 The revised Chinese-MISS (C-MISS-R) is a well validated short form.30 Two four-item subscales measure both the cognitive elements (understanding, expectations, and knowledge) and the affective elements (satisfaction, worry, and interpersonal elements) of consultations. Items are scored on a five-point Likert scale from “strongly agree” to “strongly disagree.” Higher scores reflect greater satisfaction.

A physical symptom distress measure was derived from earlier work and the literature.19, 30 For each of 14 physical symptoms, respondents indicated 0) “do not have this symptom,” or 1) “mild,” 2) “moderate,” 3) “severe,” or 4) “very severe.” All items are summed, with higher scores indicating greater physical symptom distress.

Demographic data (age, marital status, occupation, education) and medical data (surgical type, disease stage, BC history, and concurrent treatments) were gathered from patients and medical records, respectively. All measures were gathered at baseline except for the C-LOT-R (theoretically, dispositional measures remain constant irrespective of when assessed) and physical symptom distress measure, which were collected at 1 month follow-up, and medical record data were collected subsequently. CHQ12 scores at 1 month follow-up also were collected.

Data Analysis

Analyses initially examined correlations between baseline and consultation variables. Analyses of variance (ANOVAs) examined differences in CHQ12 baseline scores as a function of categorical variables, including medical variables. Then, correlations identified variables that were associated significantly with follow-up CHQ12 scores. Variation by categorical variables was examined using ANOVA. Wilcoxon rank tests, t tests, and chi-square analyses were used to compare ranked, continuous, and categorical data. Next, correlates with baseline CHQ12 scores were entered into forward-entry, stepped, multivariate, linear regression models to adjust follow-up CHQ12 score prediction. Stepped regression automates model entry criteria (F test probability; P < 0.05) and removal criteria (at P > 0.10) and assesses each variable contribution (R² change), maximizing accounted variance (adjusted R²) with the most parsimonious solution. This enables adjustment for confounding. These analyses tested Hypotheses 1–4. To identify psychosocial, demographic, and surgical predictors of psychological morbidity at 1 month follow-up, 5 regression models were calculated. Model 1 regressed CHQ12 scores on correlates of CHQ12. Significant predictors were entered into Model 2, which regressed follow-up CHQ12 scores on predictors from Model 1 and adjusted for postsurgical treatment influences. Model 3 regressed follow-up CHQ12 scores on predictors from Model 2 while adjusting for physical symptom distress. To identify predictors of physical symptom distress, Model 4 regressed physical symptom distress on treatment and baseline psychological factors. Next, Model 5 regressed follow-up CHQ12 scores on prior predictors from Models 1–4 and included lymph node status to adjust for influence of prognostic knowledge on distress. Significant collinearity (variance duplication) was inferred if condition indices were > 30 and variance proportions were > 0.5 for ≥ 2 variables.33 The inclusion and retention of individual variables in a model was based on the variable's associated R² change and judgment of the variable's overall contribution.33 Retained variables with an R² change < 0.01 were ignored for the purposes of model interpretation. All analyses used SPSS software (version 11.0).

RESULTS

At follow-up, 367 of 405 participating women did not differ by demographic or medical criteria, surgery type, or disease stage from nonparticipants. Most nonparticipants were married (82%), and 60% had no formal education. Among the participants, 33% had only a primary education, and 56% had only a secondary education. Most women (79%) were married, and ranged in age from 28 years to 79 years (mean ± standard deviation, 51.8 years ± 11.1 years). Among the women who had staging information available, 28% had Stage I BC, 49% had Stage II BC, and 18% had Stage 0 BC. Demographic and treatment information is presented in Table 1. All hospitals involved performed reconstruction simultaneously with mastectomy. Hence, no women were facing additional elective surgery.

Table 1. Sample Characteristics (N = 367 patients)

Characteristic	No. of patients (%)
SD: standard deviation; HK$: Hong Kong dollars; MRM: mastectomy; BCT: breast-conserving treatment a One United States dollar (US$1) = HK$7.8; HK$10,000 = US$1282; HK$30,000 = US$3846. b Not all women had children, so the total number was < 367 women.
Demographic information
Mean ± SD age (yrs)	51.1 ± 10.3
Marital status
Single	41 (11.2)
Married/cohabiting	275 (74.9)
Divorced/separated	25 (6.8)
Widowed	26 (7.1)
Education level
No formal education	32 (8.7)
Primary education (up to 6 years formal education)	123 (33.5)
Secondary education (Completed high school education)	170 (46.3)
Tertiary education (college/university education)	42 (11.4)
Total monthly household income (HK$)a
< $10,000	128 (37.1)
$10,001–20,000	110 (31.9)
$20,001–30,000	48 (13.9)
$30,000–40,000	30 (8.7)
> $40,000	29 (8.4)
Missing data	22 (5.9)
Living arrangement
Living alone	22 (6.0)
Living with family/significant others	345 (94.0)
Age of childrenb
< 18 Yrs	81 (26.7)
≥ 18 Yrs	222 (73.3)
Occupation
Full-time employment	130 (35.4)
Part-time employment	21 (5.7)
Retired	65 (17.7)
Housewife	105 (28.6)
Unemployed before diagnosis	23 (6.3)
Unemployed after diagnosis	23 (6.3)
Hong Kong residency
≤ 7 Yrs	9 (2.5)
> 7 Yrs	358 (97.5)
Family history of breast carcinoma
Yes	33 (9.0)
No	334 (91.0)
Medical information
Mean (± SD) mos since breast carcinoma diagnosis	1.37 ± 1.86
Mean (± SD) days since surgery	2.40 ± 1.67
Stage of disease
0	44 (12.7)
I	94 (27.1)
II	182 (52.4)
III	25 (7.2)
IV	2 (0.6)
Missing	20 (5.4)
Type of surgery
BCT	76 (20.7)
MRM	262 (71.4)
MRM and breast reconstruction	22 (6.0)
BCT and MRM	7 (1.9)
Current adjuvant therapy
Chemotherapy	107 (55.7)
Radiation therapy	25 (13.0)
Hormone therapy	66 (34.4)
Not currently receiving adjuvant therapy	169 (46.1)

Self-efficacy, optimism, and consultation satisfaction were high. Mean scores for outcome expectancies were modest, whereas mean TDM difficulties scores fell into the second lowest quartile. Physical symptom distress was slightly higher in women who were on active treatment but was quite low overall (Table 2). At follow-up, the proportions of women who reported physical symptom distress as “moderate” or worse were 45% for fatigue (mean score ± standard deviation: 1.38 ± 1.13), 41% for sleep problems (1.07 ± 1.22), 23% for pain (0.89 ± 0.94), 22% for weakness (0.75 ± 1.01), 20% for appetite change (0.65 ± 1.07), and 18% for nausea (0.59 ± 1.03). The median time from diagnosis to surgery was 25 days (mean ± standard deviation, 38.7 days ± 54 days). The mean time from decision to surgery is unknown but was less than the time from diagnosis to surgery. At follow-up, 107 women (29%) were on active chemotherapy, 73 women (20%) were awaiting chemotherapy, 21 women (6%) were unsure whether they needed chemotherapy, and 166 women (45%) were not expecting to have chemotherapy. CHQ12 scores at follow-up varied significantly according to chemotherapy status (F = 8.14; 3 degrees of freedom [df]; P < 001 [post hoc Bonferroni analysis]; women who were not expecting chemotherapy had significantly lower CHQ12 scores compared with women who were on active chemotherapy and women who were waiting for chemotherapy). Therefore, this variable comprised “active chemotherapy” in subsequent analyses.

Table 2. Means and Standard Deviations of Measures Used at Baseline and at Follow-Up

Variable	Baseline (n = 405)	Follow-up (n = 367)
Range: Maximum range of possible scores; SD: standard deviation.
Psychological distress (Chinese Health Questionnaire-12)
Mean ± SD	10.05 ± 6.42	9.38 ± 6.32
Range	0–36
Perceived self-efficacy (Generalized Self-efficacy Scale)
Mean ± SD	29.63 ± 6.01
Range	0–40
Optimism (Chinese version of the revised Life Orientation Test)
Mean ± SD		16.93 ± 2.66
Range		0–24
Patient satisfaction with the medical consultation (modified versions of the Medical Interview Satisfaction Scale)
Mean ± SD	31.28 ± 4.44	30.82 ± 4.49
Range	0–40
Treatment outcome expectancy (Breast Cancer Treatment Decision-Making Questionnaire)
Mean ± SD	21.44 ± 6.61
Range	13–65
Treatment decision-making perceived difficulties (barriers)
Mean ± SD	10.39 ± 4.23
Range	6–24
Treatment decision-making perceived difficulties (facilitators)
Mean ± SD	2.93 ± 1.26
Range	2–8
Physical symptom distress (on active treatment)
Mean ± SD		9.52 ± 7.75
Range		0–56
Physical symptom distress (no active treatment)
Mean ± SD		6.03 ± 5.44
Range		0–56

Prevalence of Psychological Morbidity

At baseline, 28% of women met the criteria for mild morbidity, and 42% of women met the criteria for moderate/severe morbidity. At 1 month follow-up, 32% women met the criteria for mild psychological morbidity, and 36% of women met the criteria for moderate/severe psychological morbidity (Table 3). Seven percent fewer women reported moderate/severe distress, and 3.5% more women reported no distress compared with baseline. To exclude the possibility of inflated distress due to contamination of CHQ12 items that addressed physical symptoms of distress (pressure in the chest, palpitations, headaches) from treatment side effects, CHQ12 scores were recalculated to exclude 3 physical symptom items. The resulting total was prorated (multiplying by 1.25) to adjust total scores. No significant changes resulted (Table 3).

Table 3. Changes in Psychological Morbidity from Baseline to Follow-Up

CH12 scorea	No of patients (%)
	A. Baseline (n = 405)	B. Baseline (n = 367)	C. Lost to follow-up	D. Follow-up (n = 367)	E. % Change (D - B)
A: total baseline sample; B: baseline sample with matching follow-up data; C: (A minus B) lost to follow-up 1 month postsurgery CHQ12: Chinese Health Questionnaire 12-item instrument. a Total number of patients with CHQ12 scores at follow-up that were lower than CHQ12 scores at baseline = 186 patients; total number of patients with CHQ12 scores at follow-up that were greater than CHQ12 scores at baseline = 143 patients; total number of patients no change in CHQ12 scores = 38 patients (Z = −2.911; P < 0.004).
CHQ12 score
Low (0–5)	120 (29.6)	106 (28.9)	14 (0.3)	119 (32.4)	13 (3.5)
Mild (6–10)	114 (28.1)	105 (28.6)	9 (0.2)	117 (31.9)	12 (3.3)
Moderate/severe (> 10)	171 (42.2)	156 (42.5)	15 (0.3)	131 (35.7)	−25 (6.8)
Prorated CHQ12 scores
Low (0–5)		108 (29.4)		125 (34.1)	17 (4.6)
Mild (6–10)		103 (28.1)		112 (30.5)	9 (2.4)
Moderate/severe (> 10)		156 (42.5)		130 (35.4)	−26 (7.1)

Although 186 women (51%) reported declines in psychological distress between baseline and follow-up, 143 women (39%) reported increases in distress during the same period (Wilcoxon z = − 2.911; P = 0.004). To determine whether women who were on active treatment were more likely to report increased CHQ12 scores, we compared the direction of change in CHQ12 scores from baseline to follow-up according to involvement in active chemotherapy. Among the women who were on active chemotherapy (107 women; 29%), 50 women (27%) showed deterioration compared with 53 women (37%) who improved, whereas 89% of 37 women with distress levels that did not change were not on chemotherapy (chi-square = 10.72; 2 df; P = 0.005). Next, we examined whether the direction of change of CHQ12 scores from baseline to follow-up was predicted by treatment after adjusting for other variables. Forward-entry, stepped logistic regression was adjusted for type of surgery, chemotherapy, TDM difficulties, negative outcome expectations, optimism, self-efficacy TDM participation incongruence, physical symptom distress, age, marital status, and education. Only physical symptom distress mildly inhibited declines in CHQ12 scores (odds ratio [OR], 0.925; 95% confidence interval [95%CI], 0.960–0.891). When physical symptom distress was removed from the analysis, chemotherapy showed similar effects, limiting CHQ12 score declines (OR, 0.506; 95%CI, 0.302–0.848).

Psychological Morbidity Predictors

Baseline morbidity.

Baseline CHQ12 scores were unrelated to type of surgery, disease stage, place of surgery, demographic factors, or degree of TDM involvement. Significant correlations emerged with perceived expectation of outcome (r = 0.506), TDM difficulties (r = 0.445), age (r = − 2.218), C-MISS-R (r = − 0.209), and GSeS (r = − 0.330). In turn, TDM difficulties were correlated with outcome expectations (0.495) and the affective (− 0.333) and cognitive (− 0.206) domains of the C-MISS-R. Multiple regression of baseline CHQ12 scores on all of the above variables retained only outcome expectation, TDM difficulties, and GSeS (adjusted R² = 0.312).

Follow-up psychological morbidity.

Follow-up CHQ12 scores were unrelated to surgery type (mean CHQ12 scores were 9.5 for BCT, 9.4 for MRM, and 8.9 for MRM+R), disease stage, hospital factors, and demographic factors. Women whose perceived TDM involvement was “as much as I desired” had significantly lower CHQ12 scores at follow-up (mean score, 8.8) than women who perceived that “I was not allowed to participate at all” (mean score, 14.7; F = 6.702; 3 df; P < 0.001; Bonferroni post hoc difference, − 5.96; P = 0.011). No underlying variable identified women who fell into the “no participation” group. Follow-up CHQ12 scores were correlated significantly with baseline CHQ12 scores (0.570), age (− 0.186), employment status (− 0.161), outcome expectations (0.434), and TDM difficulties (0.339); and, at follow-up, follow-up CH12 scores were correlated with optimism (− 0.369) active chemotherapy (0.143), and physical symptom distress (0.625).

Forward-entry, multiple regressions were adjusted for confounding factors. Model 1 regressed follow-up CHQ12 scores on significantly correlated baseline variables to test the contribution of baseline psychosocial variables to distress (Hypotheses 1 and 2). Perceived outcome expectancies and TDM difficulties predicted follow-up CHQ12 scores after 2 iterations, supporting Hypothesis 1 (Table 4). Model 2 adjusted the output from Model 1 (outcome expectancies and TDM difficulties) for the effects of active treatment by adding active chemotherapy and radiotherapy plus optimism. Outcome expectations, optimism, chemotherapy, TDM difficulties, and radiotherapy were retained in five iterations (Table 4). Model 3 adjusted the output of Model 2 for the effects of ongoing physical symptoms by including physical symptom distress at follow-up. Physical symptoms, optimism, outcome expectations, chemotherapy, and TDM difficulties all were retained in 5 iterations to account for 52% of the variance in follow-up CHQ12 scores (Table 4).

Table 4. Forward, Stepwise, Multiple Regression Models of the Chinese Health Questionnaire 12- Item Instrument (Models 1, 2, 3, and 5) and Physical Symptom Distress (Model 4) on Predictors

Modela	Step	R^2 change	Adjusted R^2	β coefficient	P value
R^2: regression coefficient; CHQ12: Chinese Health Questionnaire 12-item instrument; TDM: treatment decision making. a Excluded variables: Model 1: age, employment status, diagnosis, consultation satisfaction, TDM participation satisfaction, self-efficacy; Model 2: all entered variables were retained; Model 3: radiotherapy. b Currently active treatment. Rows in parentheses indicate variables that contributed ≤ 0.01 change to the R^2 value. Although they were significant statistically, these variables were not considered meaningful additions to the model and, thus, were not considered in the interpretation.
Model 1 (2 steps): Dependant, CHQ12 at follow-up (retained variables)
1. Outcome expectations	1	0.174	0.172	0.327	< 0.001
2. TDM difficulties	1 + 2	0.023	0.193	0.162	0.002
Model 2 (5 steps)
1. Outcome expectations	1	0.189	0.186	0.249	< 0.001
2. Optimism	1 + 2	0.069	0.254	−0.292	< 0.001
3. Chemotherapyb	1 + 2 + 3	0.020	0.272	0.129	0.004
4. TDM difficulties	1 + 2 + …4	0.016	0.287	0.151	0.003
(5. Radiotherapyb	1 + 2 + …5	0.008	0.293	−0.091	0.043)
Model 3 (5 steps)
1. Physical symptoms	1	0.397	0.395	0.607	< 0.001
2. Optimism	1 + 2	0.710	0.465	−0.179	< 0.001
3. Outcome expectations	1 + 2 + 3	0.036	0.500	0.158	< 0.001
4. Chemotherapyb	1 + 2 + …4	0.020	0.519	−0.164	< 0.001
(5. TDM difficulties	1 + 2 + …5	0.007	0.525	0.100	0.018)
Model 4 (3 steps): Dependant, physical symptom distress
1. Chemotherapyb	1	0.120	0.117	0.289	< 0.001
2. Baseline CHQ12	1 + 2	0.055	0.160	0.231	< 0.001
3. Radiotherapyb	1 + 2 + 3	0.014	0.180	0.119	0.026
Model 5 (6 steps): Dependant, follow-up CHQ12 score
1. Physical symptoms	1	0.397	0.395	0.565	< 0.001
2. Baseline CHQ12	1 + 2	0.156	0.550	0.352	< 0.001
3. Optimism	1 + 2 + 3	0.022	0.572	−0.138	< 0.001
4. Chemotherapyb	1 + 2 + 3 + 4	0.014	0.585	−0.142	< 0.001
(5. Lymph node status	1 + 2 + …5	0.011	0.600	0.106	0.004)
(6. Outcome expectation	1 + 2 + …6	0.006	0.605	0.093	0.026)

Psychological morbidity may increase symptom reporting, accounting for the high effect size of physical symptoms. Therefore, regression Model 4 examined whether psychological distress at baseline predicted follow-up physical symptom distress after adjustment for age, surgery type, and tumor stage. Physical symptom distress was predicted by current chemotherapy, baseline CHQ12 score, and radiotherapy (Table 4). Model 5 adjusted for the contribution of baseline CHQ12 score on physical symptom distress by adding baseline CHQ12 and lymph node status into the output from Model 3. Only TDM difficulty was ejected. This model accounted for 60.5% of variance. However, the contributions to R² change of lymph node status and outcome expectancies were marginal. A forced-entry regression confirmed that both these variables were nonsignificant. The amended Model 5 accounted for 58.5% of variance in follow-up CHQ12 scores. Only 25 women attended a support group, too few to enter meaningfully into the multivariate model. We therefore used a t test to compare women who were attending a support group (mean follow-up CHQ12 score = 11.36) with women who were not attending a support group (mean follow-up CHQ12 score = 9.24), but the difference in CHQ12 follow-up scores was nonsignificant (t test = − 1.62; 364 df; P = 0.106). Hypotheses 1–4 were confirmed fully, except with regard to self-efficacy.

DISCUSSION

Psychological morbidity has not been reported previously in Cantonese Chinese women after undergoing surgery for BC. The current study also was unique for its rigorous adjustment of dispositional, treatment, and prognostic variables. The results of this study showed that Chinese women experience considerable psychological morbidity, with 50% experiencing moderate-to-severe distress 1 month after surgery. Distress declined over the month after surgery in 51% of women; however, for nearly 40% of women, distress increased. Physical symptom distress was mediated by the impact of chemotherapy, maintaining psychological distress.

Prior western studies described social support,34–37 body image,38 and coping responses34, 37–39 as predictors of adjustment after BC surgery. Consultation factors31, 40 seldom are described. Chinese women show similarities to western women regarding TDM and expectations.31 Chinese women with BC report fewer TDM difficulties when their preferred and actual TDM involvement match.21 TDM difficulties and outcome expectancies strongly predicted postoperative and follow-up psychological morbidity, echoing a report that physicians' attitudes and willingness to involve American women in TDM modulate adjustment.40 Because that study involved data collection 6 months after consultation,40 significant recall bias was likely. Our current study involved just over 1 month's delay between consultation and data collection, but the results remain vulnerable to recall bias. Because the two studies are congruent, our confidence is strengthened in the veracity of the data.

The current results also advance our understanding that, after adjustment for disposition and coping efficacy, consultations mediated by TDM-related factors can exacerbate baseline distress, in turn, mediating adjustment 1 month postoperatively. Outcome expectancy among American women predicts adjustment at 3 months after breast surgery.31 We have illustrated similar effects in Chinese women. Although retrospective TDM assessment weakens our inferences, prospective studies of real-world TDM without significant contamination are next to impossible.30

Physical symptom distress, which accounted for most of the variance in follow-up CHQ12 scores, was predicted by active treatment and baseline CHQ12 scores. Chemotherapy paradoxically enhanced physical symptom distress but lowered follow-up psychological morbidity. Chinese women may believe that they are receiving additional treatment, thereby gaining reassurance from chemotherapy while experiencing the side-effects.19 Therefore, symptoms are worse, but worry is less. We did not find similar reports on western patients. Symptom experience is a function of perceptions of the causes, attributions, and, hence, meanings of the experienced symptoms.41 We speculate that, in the absence of active treatment, women may attribute symptoms and sensations to disease progress. Understandably, this increases distress. However, chemotherapy's notorious side effects prompts the attributing of symptoms to treatment rather than to disease in women receiving chemotherapy. Hence, the meaning of symptoms does not invoke the threat of disease progression. These data confirm the phenomenological experiences of Chinese women recovering from BC13 and reports on patients with lymphoma in the United States.42 Hence, we suggest that physical symptom distress is the strongest predictor of distress in our multivariate models, not because it reflects direct treatment side effects but, rather, because it reflects uncertainty about what the symptoms mean. The specter of disease progression despite surgery and, by extension, of further life disruption and pain looms large at this time for Chinese, as we suspect it does for most women facing BC.

This study was limited methodologically by retrospective data collection for consultation, outcome expectations, and TDM. Bias can arise from forgetting and contamination by intervening events or outcomes. However, > 80% of data were collected within 3 days after surgery, and 100% of data were collected within 5 days after surgery. Women had not yet viewed their surgical wounds, which limited the effects of regret or disappointment regarding appearance. Recall time from consultation to interview was < 33 days on average. The second major limitation was in cross-sectional interactions between physical symptom distress and follow-up CHQ12 scores, which cannot be disentangled here.

Baseline psychological morbidity reflected TDM difficulties and, thus, treatment outcome uncertainties and had an impact on subsequent psychological morbidity. No prior studies reported that TDM could predict psychological morbidity in Chinese women with BC. Although optimism also predicted follow-up psychological morbidity, it is a well established predictor35, 41 that benefits Chinese and other groups.13 TDM and outcome-expectation effects were independent of optimism. In follow-up, all women had been told their lymph node status; however, this made only a very minor addition to the model. Accounting for ≈ 60% of variance in psychological distress is a significant advance on earlier studies. Whereas we emphasize consultation and TDM factors, after adjustment, among Chinese women with BC, the data also extend our understanding of the development of distress in women, bringing together for the first time both consultation and cognitive elements reported in the western literature and illustrating intimate relations between uncertainty, attributions and negative affect, and consultation. Chinese women differed from their western counterparts, in that marital status effects were absent, suggesting different expressions or roles for social support among Chinese communities. The fact that most elements of western studies were upheld suggests that, for the most part, making sense of BC distress can be influenced strongly by current tertiary approaches to BC management, consultation practices, and the management of uncertainty. In this report, we began by emphasizing uncertainty in deciding BC treatment. The current results uphold this view and emphasize the ongoing role of uncertainty in psychological distress thereafter.

These data suggest clinicians should assess preferred TDM involvement at the initial consultation in Chinese women with BC. Most Chinese women prefer shared TDM, one-third prefer to decide alone or with their family, and others prefer having the surgeon decide.21 Women may need help or support with deciding on treatment. Decision-making support through treatment recommendation,15 accurate information about appearance outcomes,11 and information about expected symptoms after surgery is desirable. Psychological distress evaluations at 1 month should be sensitive to interventions that provide personalized support around expected symptoms and decision making and that prepare patients for physical symptoms. Minimizing opportunities for misattribution helps reduce unnecessary anxiety. Interactions with other women who have completed BC treatments successfully are likely to be particularly beneficial.22 Chinese women who have significant difficulties in TDM should be considered at high risk for developing later psychological distress.

Acknowledgements

The authors thank the participating women and the Chiefs of Service at the participating hospitals.