A total of 9331 patients completed the MWC questionnaire at least once during the study period. After excluding patients diagnosed with cancer more than 5 years prior to their initial assessment, 6939 patients were included in the analysis. The median number of assessments per patient was 2 (range: 1–12). About one third of the sample completed one questionnaire (n = 2258 (32.5%)). Over half the sample (n = 3585; 51.7%) completed two to four, and 15.8% (n = 1096) completed five or more questionnaires. The median time from diagnosis to the initial assessment was 12 months (interquartile range: 2.9–29.5).

The mean age at first assessment was 63.3 years (SD = 13.1). Of the included patients, 3005 (43.3%) were female, and 3015 (43.5%) identified as Hispanic. Most patients were insured (n = 6785, 97.8%) and partnered (n = 4346, 62.6%). Hematologic malignancies were the most common cancer type (n = 1802, 26%), followed by prostate (n = 1428, 20.6%) and GI (n = 1418, 20.4%). The average time since diagnosis was 18 months (SD = 17.1), and nearly one third (n = 2054, 29.6%) had metastatic disease at the first assessment. Additional characteristics, including cancer-specific breakdowns, are summarized in Table 1. PRO score categories at first assessment indicate that patients with prostate cancers consistently reported the highest percentages of “normal” levels for anxiety, depression, and fatigue. Conversely, GYN and lung cancer patients exhibited a higher prevalence of moderate or severe symptoms across most PROs (Table 2). Participants who completed multiple assessments were more likely to be female and non-Hispanic Black, had higher Charlson Comorbidity Index scores, and were more frequently receiving treatments, including chemotherapy, immunotherapy, or hormonal therapy compared to those who completed only one assessment (Online Resource Table S1). Conversely, single-assessment participants were more likely to have metastatic disease, higher pain interference, and a more recent cancer diagnosis (Online Resource Table S1).

The final models and results are summarized in Table 3, and the predicted means of each PRO across time by cancer type are shown in Online Resource Table S2. Figure 1 illustrates the trajectory of each PRO by cancer type.

a–f Trajectory of PRO by domain

Anxiety scores varied significantly across patients at cancer diagnosis (random intercept = 54.19, p < 0.001). Having a partner (B = −1.06, p < 0.001) was associated with lower anxiety, while non-Hispanic Black patients (B = −2.29, p < 0.001) and those identifying as non-Hispanic Other (B = −1.83, p < 0.0001) reported lower anxiety than non-Hispanic Whites at cancer diagnosis. A greater number of comorbidities (B = 0.27, p < 0.001) was associated with increased anxiety. Receiving hormonal therapy (B = 1.12, p < 0.001) or surgery (B = 1.44, p < 0.001) at the time of assessment was associated with higher anxiety.

Overall, anxiety scores decreased over time (fixed III test p < 0.001) (Fig. 1a), with significant variation in individual rates of change (random-slope variance = 0.01, p = 0.002). However, the rate of change did not differ significantly by cancer type (interaction: p = 0.178). When examining cancer types, compared to prostate cancer, patients with all other cancer types except head and neck reported higher anxiety at diagnosis (all comparisons were significant at p < 0.008). Specifically, predicted anxiety scores remained significantly elevated for patients with GYN cancer than those with GI through 24 months, head and neck through 36 months, and prostate cancer through 60 months (Table S2). Conversely, patients with prostate cancer consistently reported lower anxiety scores than patients with breast, GYN, GI, hematologic, and lung cancers from diagnosis through the first 48 months (all comparisons were significant at p < 0.008).

Depression scores varied significantly across patients at cancer diagnosis (random intercept = 54.27, p < 0.0001). Having a partner (B = −1.48, p < 0.001) was associated with lower depression, while Hispanic, non-Hispanic Black, and non-Hispanic Other patients reported lower depression than non-Hispanic Whites at cancer diagnosis (B = −1.90, p = 0.001). A greater number of comorbidities was associated with higher depression (B = 0.30, p < 0.001). Receiving hormonal therapy at the time of assessment was associated with higher depression levels (B = 0.87, p = 0.002).

Depression scores decreased over time overall (fixed III test p = 0.008), with significant variation in individual rates of change (random slope variance = 0.01, p = 0.04). However, the rate of change was similar across cancer types (interaction: p = 0.713). When examining cancer types, patients with breast, GYN, and hematologic cancers reported significantly higher depression at diagnosis than those with prostate cancer (all p’s < 0.005). Predicted depression scores indicated that the increased depressive burden for GYN and hematologic cancer patients persisted, remaining significantly higher than prostate cancer through the first 48 months (Table S2). Breast cancer patients exhibited significantly higher depression scores than those with prostate cancer early in the survivorship period (first 24 months) (all p’s < 0.008).

At cancer diagnosis, pain interference scores varied significantly across patients (random intercept = 55.49, p < 0.001). Having a partner (B = −0.99, p < 0.001) was associated with lower pain interference. Higher pain interference scores at cancer diagnosis were significantly associated with older age (B = 0.04, p < 0.001) and comorbidities (B = 0.40, p < 0.001). Receiving chemotherapy (B = 0.67, p = 0.002) or immunotherapy (B = 1.48, p < 0.001) and undergoing surgery (B = 8.39, p < 0.001) at the time of assessment were significant predictors of higher pain interference scores.

Pain interference scores changed over time (fixed III test p = 0.007), with significant variation in individual rates of change (random slope variance = 0.01, p = 0.04). However, the rate of change was similar across cancer types (interaction: p = 0.289). When examining cancer types, compared to prostate cancer patients, patients with GYN, GI, hematologic, or lung cancer reported higher levels of pain interference at cancer diagnosis. Specifically, predicted mean scores showed that patients with GYN and hematologic cancers had the highest levels of pain interference across time, while prostate cancer patients reported the lowest pain interference across time (Table S2).

At cancer diagnosis, fatigue scores varied significantly across the patients (random intercept = 64.53, p < 0.001). Having a partner (B = −1.07, p < 0.001) was associated with lower fatigue, while Hispanic patients reported lower fatigue than non-Hispanic Whites (B = −2.49, p < 0.001). A greater number of comorbidities at cancer diagnosis (B = 0.40, p < 0.001) and metastatic disease (B = 0.47, p < 0.001) were associated with higher fatigue. Receiving chemotherapy (B = 1.92, p < 0.001) and undergoing surgery (B = 2.12, p < 0.001) at the time of assessment were associated with higher fatigue.

Overall, fatigue scores decreased over time (fixed III test p < 0.001), with no significant variation in individual rates of change (random slope variance = 0.00, p = 0.478). When examining cancer types, patients with prostate cancer reported the lowest fatigue levels at cancer diagnosis. Furthermore, the rate of change was different across cancer types. Compared to patients with prostate cancer, those with head and neck (B = −0.07, p = 0.006) and hematologic malignancies (B = −0.07, p = 0.001) reported an improvement in fatigue over time. Predicted fatigue scores showed that patients with GYN and hematologic malignancies maintained significantly higher fatigue than prostate cancer patients through the first 48 months (Fig. 1d; Table S2). Patients with head and neck cancer showed significantly lower predicted fatigue scores than those with GYN and hematologic cancer from diagnosis through 60 months (all comparisons were significant at p < 0.008).

Physical function scores varied significantly across patients at cancer diagnosis (random intercept = 70.96, p < 0.001). Older age (B = −0.22), a higher number of comorbidities (B = −0.63), and metastatic disease (B = −0.95) were associated with significantly lower physical function scores at cancer diagnosis. Receiving chemotherapy (B = −1.63), immunotherapy (B = −0.92), and undergoing surgery at the time of assessment (B = −9.21) were associated with significantly lower physical function scores (all p’s ≤ 0.001). Conversely, having a partner (B = 1.05, p < 0.001) was associated with significantly higher physical function scores.

Overall, physical function scores improved over time (fixed III test p < 0.001), with significant variation in individual rates of change (random slope variance = 0.02, p < 0.001). When examining cancer types, compared to patients with prostate cancer, all other cancer types except head and neck reported lower physical function at cancer diagnosis. Patients with hematologic malignancies faced the greatest burden at cancer diagnosis; however, they showed significant greater improvements in physical functioning over time (B = 0.07, p = 0.0003). Predicted mean scores indicated that patients with prostate cancer exhibited significantly higher physical function scores through 60 months compared to those with breast, GI, hematologic, and lung cancer (Table S2). Similarly, patients with head and neck cancer had significantly higher physical function scores compared to those with breast, GI, hematologic, and lung cancer from 12 months onward. In contrast, patients with GYN cancer consistently exhibited lower physical function scores across time than patients with prostate and head and neck cancer (all comparisons were significant at p < 0.008).

HRQOL scores varied significantly across patients at cancer diagnosis (random intercept = 18.53, p < 0.001). A higher number of comorbidities (B = −0.28) and metastatic disease (B = −0.67) were associated with lower HRQOL at cancer diagnosis. Receiving chemotherapy (B = −0.80), hormonal therapy (B = −0.60), or surgery (B = −1.18) at the time of assessment was associated with lower HRQOL scores (p’s < 0.001). Having a partner (B = 0.72, p < 0.001) was associated with higher HRQOL scores at cancer diagnosis.

Overall, HRQOL scores significantly change over time (fixed III test p < 0.001). However, the rate of change was similar across cancer types (interaction: p = 0.170). When examining cancer types, compared to prostate cancer, all other cancer types except breast reported lower HRQOL at cancer diagnosis (all p’s < 0.001). Although not statistically significant, patients with head and neck cancer showed improvements in HRQOL over time compared to those with prostate cancer (B = 0.03, p = 0.026) (Fig. 1f). Prostate cancer patients had significantly higher predicted HRQOL scores than those with GYN, GI, hematologic, and lung cancer through the first 36 months (Table S2). Similarly, patients with head and neck cancer had significantly higher HRQOL scores compared to those with hematologic malignancies from diagnosis through the first 48 months. By 60 months, there are no differences between cancer types, suggesting a long-term stabilization in HRQOL.