Course and predictors of supportive care needs among Mexican breast cancer patients: A longitudinal study

Abstract Objective This study examined the course and predictors of supportive care needs among Mexican breast cancer patients for different cancer treatment trajectories. Methods Data from 172 (66.4% response rate) patients were considered in this observational longitudinal study. Participants were measured after diagnosis, neoadjuvant treatment, surgery, adjuvant treatment, and the first post‐treatment follow‐up visit. Psychological, Health System and Information, Physical and Daily Living, Patient Care and Support, Sexual, and Additional care needs were measured with the Supportive Care Needs Survey (SCNS‐SF34). Linear mixed models with maximum‐likelihood estimation were computed. Results The course of supportive care needs was similar across the different cancer treatment trajectories. Supportive care needs declined significantly from diagnosis to the first post‐treatment follow‐up visit. Health System and Information care needs were the highest needs over time. Depressive symptoms and time since diagnosis were the most consistent predictors of changes in course of supportive care needs of these patients. Conclusions Health system and information care needs of Mexican breast cancer patients need to be addressed with priority because these needs are the least met. Furthermore, patients with high depressive symptoms at the start of the disease trajectory have greater needs for supportive care throughout the disease trajectory.

research among Asian and European breast cancer patients showed that supportive care needs may change during active treatment, 6,7 and in the survivorship phase. 8 Besides, cross-sectional research showed differences in unmet care needs between Asian and European breast cancer patients. 9 A longitudinal study among Taiwanese breast cancer patients in early stages of the disease and undergoing active treatment showed that supportive care needs decreased from diagnosis up to 3-month follow-up post-diagnosis. 6 By contrast, another study with Chinese breast cancer patients in more advanced stages of the disease receiving chemotherapy (baseline) and who were followed up to 12 Additionally, these studies did not distinguish between the different treatment trajectories, included only patients in an advanced stage of the disease 7 or in the survivorship phase, 8

| Design
We conducted an observational longitudinal study with measurements scheduled after diagnosis, after finishing each treatment modality, and at the first post-treatment follow-up visit. Breast cancer patients differ in treatment trajectories depending on cancer stage and several prognostic factors. 11 Overall, patients with advanced cancer stages and worse prognostic factors follow more intensive treatment trajectories compared with patients with earlier cancer stages or better prognostic factors. Thus, treatment trajectories are intertwined with cancer stage. Based on the treatment trajectories that the patients in our study followed, we identified 3 groups. Group A are the patients who followed surgery only, after the diagnosis.
Group B are the patients who followed surgery and adjuvant treatment. Group C are the patients who followed neoadjuvant treatment, surgery, and adjuvant treatment. The assessments were conducted after diagnosis but before surgery (T1); after end of neoadjuvant treatment (T2); after surgery and before start of adjuvant treatment (T3); at the end of the adjuvant treatment (either chemotherapy, radiotherapy or the combination of both) (T4); at first post-treatment follow-up appointment (T5). Depending on these treatment trajectories, patients had 3 (group A), 4 (group B), or 5 measurements (group C). Some patients within each group were following hormone therapy, but this treatment was not taken into account in the study due to its long duration.  needs. For this study, we added an extra dimension that we labeled as "Additional needs" (5 items) from the long version of the SCNS 14 and refers to financial and practical difficulties. Thus, we used 39 items to measure the patients' level of need for help over 2 weeks preceding the interview. Details on the adaptation of the instrument to the Spanish version can be found elsewhere. 12 The instrument employs a 5-point Likert response scale to rate the intensity of each need, that is, 1 = Not applicable, 2 = Satisfied, 3 = Low need, 4 = Moderate need, and 5 = High need. A total score for each dimension was computed using standardized scores, which ranged from 0 to 100. Higher scores reflected higher supportive care needs. 15 Cronbach alphas for the subscales at baseline ranged from .69 to .95.

| Participants and procedure
Anxiety symptoms were measured at baseline with the short form of the Spielberg State and Trait Anxiety Inventory. 16 This version employs 6 items from the state subscale of the long original version.
We used the equivalent 6 items from the Spanish version of the instrument. 17 A total score is obtained summing all the items after the negatives items have been properly transformed. Higher scores indicate higher levels of anxiety (1 = not at all, 4 = very much).
Cronbach alpha for the scale was.81.  19 We used the 3 symptoms scales: systemic therapy side effects (7 items), breast symptoms (4 items), and arm symptoms (3 items). The response scale ranges from 1 = not at all to 4 = very much.

FIGURE 1 Flow chart of the patients' follow-up
Higher scores represent higher physical symptoms. Cronbach alphas for the scales ranged from .60 to.71.
At baseline, we also collected data on age, number of children, marital status, education level, work status, and life events during the last 3 months and whether patients were receiving psychological aid at the moment of the interview. Marital status was classified into with partner (married/living together/in a relationship) and without partner (single/ widow/divorced); education level was classified into high (bachelor/ postgraduate studies), middle (secondary/technical high school), low (without studies/primary); work status was classified into housewife or employed, and life events into yes or not. We also collected information on comorbidities, type of treatment and cancer stage.

| Statistical analyses
We computed descriptive statistics of the sample characteristics per cancer treatment group. ANOVAs and chi-square tests were run to compare the baseline characteristics of each group. To analyze the longitudinal course of supportive care needs, we computed linear mixed models with maximum-likelihood estimation for each subscale separately, including group, time, and their interaction. Time is the number of days since diagnosis, and it was treated as a continuous variable because the lapse between each measuring point varied for each patient. We computed both, models with random intercept and random slope, and models with random intercept only. We used Akaike information criterion (AIC) and Bayesian information criterion (BIC) to select the best possible model. Lower values of AIC/BIC indexes were considered as indicative of a better model. According to these criteria, only the intercept was considered to be random in models for psychological, health system and information, physical and daily living, patient care and support, and sexual dimensions; whereas a random intercept and random slope were considered for additional care needs dimension. Subsequently, predictors of the supportive care needs' course were identified through univariate analyses.
Those sociodemographic, physical, and psychological variables (anxiety and depressive symptoms at baseline), which were significantly related to specific supportive care needs subscale, in at least 2 measurement points were included. We recomputed linear mixed models  Cancer stage was significantly associated to group classification, most patients with cancer stage III or IV were allocated to group C. Further details on the sample's characteristics are in supporting   Figures 2C1-2C3). We observed a decrease in Patient Care and Support needs for all groups, but it was statistically significant only for groups B and C (Figures 2D1-2D3). The course of Sexual care needs for groups A and B was low without significant changes over time, but there was a significant small decrease over time for group C ( Figures 2E1-2E3). The Additional care needs showed a significant decline pattern in all 3 groups (Figures 2F1-2F3).

| Course of supportive care needs
The interaction between days since diagnosis (time) and treatment trajectory (group) was not significant for any of the supportive care needs dimensions, suggesting that the changes on the course of supportive care needs were similar across the cancer treatment trajectories. However, we observed a significant difference in the level of Patient Care and Support dimension at T1 between the treatment trajectories of group A and group C ( Table 1). On average, patients from group A at T1 started with lower levels in this dimension ( Figure 2D1), compared with patients from group C ( Figure 2D3, P = 0.02).

| Predictors of supportive care needs
After adjusting linear mixed models by potential predictors, results showed that, in general, patients with higher levels of depressive or anxiety symptoms after diagnosis, those who received psychological aid at T1, and those with higher systemic therapy side effects after surgery showed higher care needs in specific domains over time.
Specifically, older patients indicated lower Sexual care needs over time. Also, patients with a partner and those who followed surgery  Table 2, which shows the significance and size of the fixed effects. In general, supportive care needs of the patients were low and decreased over time, which is in line with previous studies among Chinese, Taiwanese, and French breast cancer patients. [6][7][8] This might suggest that patients' needs are met; either they receive the care they require from the health system or they manage themselves to get the help they need outside, particularly within a collectivistic culture like Mexico, where social relationships and attachment between family members is highly present. Health System and Information care needs were the highest throughout time. This is consistent with previous studies among Asian patients, 6,7 but in contrast with studies among Caucasian patients where Psychological needs were also high. 8,20 This

| Study limitations
The results of this study should be interpreted considering some limitations. In this study, physical and psychological symptoms were assessed at 1 point in time. Physical symptoms were measured after surgery, and it might be that for patients who received neoadjuvant treatment, physical symptoms started to exhibit earlier in the disease trajectory. Also, the relatively small sample size used might have prevented us from identifying a significant interaction between time and group. A higher attrition was observed among the patients in advanced cancer stage or with lower education, which is common in longitudinal studies involving (low-middle income) cancer patients. 29,30 Although there are studies indicating that selective attrition does not always affect the estimates of associations between variables, 31