Hospital care patterns among patients with psychotic-disorder coding and metastatic/secondary malignant neoplasm coding: A nationwide administrative cohort study in Spain.
Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer · 2026
Abstract originale
PURPOSE: Patients with psychotic disorders may experience differences in cancer care, but evidence on hospital supportive and palliative care patterns in advanced cancer remains limited. We examined whether psychotic-disorder coding was associated with hospital palliative-care coding and selected hospital treatment-intensity markers among patients with metastatic/secondary malignant neoplasm coding in a Spanish hospital oncology extract. METHODS: We conducted a nationwide retrospective cohort study using Spanish hospital administrative data from 2021 to 2024. The source extract was a requested oncology administrative hospital extract, composed overwhelmingly of episodes with malignant cancer recorded as the principal diagnosis and also containing a small subset of palliative-care principal-diagnosis episodes with malignant neoplasm/metastatic coding in secondary fields. Eligible patients had secondary malignant neoplasm/metastatic disease coding in secondary diagnosis fields (C77-C79). The first metastatic-coded episode observed during 2021-2024 was treated as an administrative index episode and not necessarily as the onset of metastatic disease. The unit of analysis was the patient. For each patient, a metastatic care window was defined from the first metastatic-coded hospital episode to the last available episode. Psychotic-disorder coding was defined primarily as F20-F29 within this window; a temporal audit examined F20-F29 coding before or at the index episode. Outcomes were considered present if coded at least once during the metastatic window. Multivariable logistic regression adjusted for age, sex, year, tumor group, autonomous community, hospital service, and approximate non-oncologic Charlson-based comorbidity measured at the first metastatic-coded episode. RESULTS: The cohort included 272,809 patients, of whom 1989 had psychotic-disorder coding within the metastatic care window (0.73%). A temporal audit identified 1938 patients with F20-F29 coding before or at the first metastatic-coded episode in the observed extract; 1889 patients met both definitions, 100 had F20-F29 coding in the metastatic care window only after the index episode, and 49 had F20-F29 coding before or at the index episode but not within the metastatic care window. Psychotic-disorder coding was associated with higher odds of hospital palliative-care coding (OR 1.35, 95% CI 1.19-1.52) and mechanical ventilation (OR 1.44, 95% CI 1.15-1.82). In-hospital chemotherapy procedure coding showed lower adjusted odds among patients with psychotic-disorder coding (OR 0.73, 95% CI 0.59-0.91). Marginal adjusted probabilities for palliative-care coding were 20.6% with psychotic-disorder coding versus 17.0% without psychotic-disorder coding. Corresponding probabilities were 4.5% versus 5.8% for in-hospital chemotherapy procedure coding and 3.6% versus 2.6% for mechanical ventilation. CONCLUSION: In this nationwide hospital administrative oncology extract, psychotic-disorder coding was associated with distinct hospital coding patterns after metastatic/secondary malignant neoplasm coding. The findings should be interpreted as hospital-coded associations and not as direct measures of appropriateness, treatment quality, outpatient systemic therapy, or specialist palliative-care access.
Parole chiave
Leggi il testo integrale originale
Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.
Hospital care patterns among patients with psychotic-disorder coding and metastatic/secondary malignant neoplasm coding: A nationwide administrative cohort study in Spain
Purpose
Patients with psychotic disorders may experience differences in cancer care, but evidence on hospital supportive and palliative care patterns in advanced cancer remains limited. We examined whether psychotic-disorder coding was associated with hospital palliative-care coding and selected hospital treatment-intensity markers among patients with metastatic/secondary malignant neoplasm coding in a Spanish hospital oncology extract.
Methods
We conducted a nationwide retrospective cohort study using Spanish hospital administrative data from 2021 to 2024. The source extract was a requested oncology administrative hospital extract, composed overwhelmingly of episodes with malignant cancer recorded as the principal diagnosis and also containing a small subset of palliative-care principal-diagnosis episodes with malignant neoplasm/metastatic coding in secondary fields. Eligible patients had secondary malignant neoplasm/metastatic disease coding in secondary diagnosis fields (C77-C79). The first metastatic-coded episode observed during 2021–2024 was treated as an administrative index episode and not necessarily as the onset of metastatic disease. The unit of analysis was the patient. For each patient, a metastatic care window was defined from the first metastatic-coded hospital episode to the last available episode. Psychotic-disorder coding was defined primarily as F20-F29 within this window; a temporal audit examined F20-F29 coding before or at the index episode. Outcomes were considered present if coded at least once during the metastatic window. Multivariable logistic regression adjusted for age, sex, year, tumor group, autonomous community, hospital service, and approximate non-oncologic Charlson-based comorbidity measured at the first metastatic-coded episode.
Results
The cohort included 272,809 patients, of whom 1989 had psychotic-disorder coding within the metastatic care window (0.73%). A temporal audit identified 1938 patients with F20-F29 coding before or at the first metastatic-coded episode in the observed extract; 1889 patients met both definitions, 100 had F20-F29 coding in the metastatic care window only after the index episode, and 49 had F20-F29 coding before or at the index episode but not within the metastatic care window. Psychotic-disorder coding was associated with higher odds of hospital palliative-care coding (OR 1.35, 95% CI 1.19–1.52) and mechanical ventilation (OR 1.44, 95% CI 1.15–1.82). In-hospital chemotherapy procedure coding showed lower adjusted odds among patients with psychotic-disorder coding (OR 0.73, 95% CI 0.59–0.91). Marginal adjusted probabilities for palliative-care coding were 20.6% with psychotic-disorder coding versus 17.0% without psychotic-disorder coding. Corresponding probabilities were 4.5% versus 5.8% for in-hospital chemotherapy procedure coding and 3.6% versus 2.6% for mechanical ventilation.
Conclusion
In this nationwide hospital administrative oncology extract, psychotic-disorder coding was associated with distinct hospital coding patterns after metastatic/secondary malignant neoplasm coding. The findings should be interpreted as hospital-coded associations and not as direct measures of appropriateness, treatment quality, outpatient systemic therapy, or specialist palliative-care access.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1007/s00520-026-11440-0.
Introduction
People living with psychotic disorders have higher morbidity and mortality and may face structural barriers across the cancer care continuum [1, 2]. In oncology, these barriers can operate through delayed diagnosis, fragmented follow-up, differences in treatment decision-making, comorbidity burden, communication challenges, social vulnerability, and clinical uncertainty when psychiatric symptoms coexist with advanced disease [3–5].
Supportive care is central to the management of metastatic cancer. It includes symptom control, palliative care integration, management of treatment complications, and decisions about potentially intensive interventions. Metastatic disease is the clinical setting in which supportive and palliative care decisions are especially relevant and more consistently documented in hospital records. However, less is known about whether patients with psychotic disorders have different patterns of hospital-based supportive, palliative, and oncologic care once metastatic/secondary malignant neoplasm coding is present. Previous reviews have emphasized that palliative care for people with severe or persistent mental illness, including schizophrenia spectrum disorders, remains underdeveloped and inconsistently integrated across care systems [6–12].
This question is clinically relevant because differences in care patterns may reflect appropriate individualized decisions, differences in clinical status, barriers to access, or inequities in treatment pathways. Population-based studies have reported excess cancer mortality and treatment disparities among people with mental illness, including schizophrenia and other severe mental disorders [13–19]. Administrative hospital data cannot adjudicate appropriateness at the individual level, but they can identify population-level signals that require clinical and policy attention.
We therefore conducted a nationwide retrospective cohort study using a Spanish hospital oncology administrative extract to examine whether psychotic-disorder coding was associated with hospital palliative-care coding among patients with metastatic/secondary malignant neoplasm coding. As exploratory secondary objectives, we evaluated selected hospital treatment-intensity markers during the metastatic care window, including intensive care unit admission, in-hospital chemotherapy procedure coding, mechanical ventilation, transfusion, and tracheostomy.
Methods
Study design and data source
We performed a retrospective cohort study using Spanish hospital administrative data from 2021 to 2024, called RAE-CMBD1. The source database was a requested oncology administrative extract for the study period, composed overwhelmingly of episodes with malignant cancer recorded as the principal diagnosis and including a small subset of palliative-care principal-diagnosis episodes with malignant neoplasm/metastatic coding in secondary fields. Diagnoses and procedures were coded using ICD-10-CM/PCS-derived Spanish coding fields. Patients were linked using the recoded patient identifier supplied in the RAE-CMBD extract. We treated this official linkage variable as the longitudinal patient key across observed episodes; the extract did not include documentation allowing us to independently audit the identifier-generation algorithm.
The extract does not represent all people with metastatic cancer in Spain or all hospitalizations experienced by those patients. It contains hospital episodes included in the requested oncology extract, which was composed overwhelmingly of episodes where malignant cancer was recorded as the principal diagnosis. Admissions primarily coded for infection, respiratory failure, psychiatric deterioration, or other non-cancer principal diagnoses may therefore be absent.
No hospital identifier or death date was available in the analysis extract. The extract included autonomous community and discharge-status fields, but not a validated date of death.
The study is reported in accordance with the STROBE Statement [20].
Study population
The unit of analysis was the patient. Patients were identified using the recoded patient identifier available in the database. We included patients with metastatic/secondary malignant neoplasm coding, defined as at least one hospital episode with a secondary diagnosis code C77, C78, or C79. Secondary diagnosis fields were used because the source extract was an oncology administrative extract in which the principal diagnosis generally captured the primary tumor or, for a small subset, palliative-care coding; secondary diagnosis fields captured secondary malignant neoplasm coding while preserving the principal-diagnosis field for tumor grouping.
Patients whose administrative index episode had Z51.5 as the principal diagnosis were retained only when C77-C79 metastatic/secondary malignant neoplasm coding was present in secondary diagnosis fields; these patients represented 4814 patients (1.8% of the analytic cohort).
Codes C77-C79 were used to define the primary cohort. Because C77 may include lymph-node secondary malignant neoplasm coding that does not always correspond to distant stage IV disease, we performed a sensitivity analysis restricted to C78-C79.
For each patient, we defined a metastatic care window starting at the first hospital episode with secondary malignant neoplasm/metastatic disease codes and ending at the last available hospital episode during the study period. The first metastatic-coded episode observed in 2021–2024 was an administrative index episode. Patients may therefore have had newly documented or prevalent metastatic disease.
The metastatic care window was used to align the exposure and outcomes with the period in which advanced cancer was documented in hospital records. We considered outcomes present if they were coded at least once during this window because the events of interest may occur in different admissions. Restricting the analysis to the first metastatic-coded episode would have underestimated patient-level exposure to these forms of care and would have treated subsequent hospital information as irrelevant despite the longitudinal nature of metastatic cancer care.
Exposure
The primary exposure was psychotic-disorder coding, defined as any secondary diagnosis code F20-F29 within the metastatic care window. These codes include schizophrenia, schizotypal disorder, persistent delusional disorders, acute and transient psychotic disorders, schizoaffective disorders, and other non-organic psychotic disorders. The exposure was assigned at patient level to avoid treating multiple admissions from the same person as independent observations.
Because this definition does not by itself establish temporal precedence, we audited F20-F29 coding before or at the first metastatic-coded episode in the observed extract. We also described the distribution of individual F20-F29 diagnostic groups and specified sensitivity analyses focused on schizophrenia/schizoaffective disorder coding.
Outcomes
The primary outcome was hospital palliative-care coding during the metastatic care window, defined by code Z51.5/Z515 in any diagnosis field. This outcome indicates hospital palliative-care coding and does not identify specialist palliative-care referral, timing, content, intensity, quality, or preference-concordance.
Secondary outcomes, all measured during the metastatic care window, were intensive care unit admission, in-hospital chemotherapy procedure coding, mechanical ventilation, transfusion, and tracheostomy. Intensive care unit admission was based on the administrative ICU field. Procedure-based outcomes were identified using prespecified ICD-10-PCS-derived procedure algorithms. In-hospital chemotherapy procedure coding does not capture outpatient chemotherapy, day-hospital treatment, immunotherapy, targeted therapy, oral anticancer therapy, or all systemic anticancer treatment. Detailed definitions are provided in the supplementary appendix.
Covariates
Models adjusted for age, sex, calendar year, tumor group based on the primary diagnosis, autonomous community, hospital service, and approximate non-oncologic Charlson-based comorbidity score, all measured at the first metastatic-coded episode. These covariates were selected a priori because they capture demographic factors, cancer type, temporal trends, territorial and organizational differences in hospital care, and non-oncologic comorbidity burden. Oncologic Charlson components were not included in the comorbidity score because the cohort was defined by metastatic/secondary malignant neoplasm coding [21, 22].
Statistical analysis
Patient characteristics and hospital outcomes were summarized overall and by psychotic-disorder coding status. Categorical variables are presented as counts and percentages, and continuous variables as mean and standard deviation.
We fitted multivariable logistic regression models for each outcome. The main parameter of interest was the adjusted odds ratio for psychotic-disorder coding. The primary analysis used the full patient cohort and outcomes defined within the metastatic care window. Categorical covariates were included using indicator variables. Patients with missing age were excluded from complete-case model fitting; missing or unavailable categorical values were retained as explicit missing/not-recorded categories where applicable.
We estimated marginal adjusted probabilities by setting psychotic-disorder coding to present and absent for all patients in the fitted model and averaging predicted probabilities. Adjusted risk differences were calculated as the difference between these marginal probabilities. Two-sided p values and 95% confidence intervals are reported. For secondary outcomes, interpretation prioritized effect size, direction, and confidence intervals rather than p values alone, and these outcomes were treated as exploratory.
We performed sensitivity analyses restricted to patients with C78-C79 coding, using psychotic-disorder coding before or at the index episode, and omitting hospital service. Hospital service at the first metastatic-coded episode was included in the main model to account for baseline hospital context, but it may partly represent the pathway under study; therefore, we added a sensitivity analysis omitting hospital service. Hospital-level cluster-robust standard errors could not be estimated because no hospital identifier was available. Models adjusted for autonomous community, but residual hospital-level clustering and coding-practice variation remain possible.
All reported logistic regression models converged. Sparse exploratory outcomes, particularly tracheostomy, were interpreted cautiously because low event counts can make estimates less stable.
Results
Cohort
The source dataset contained 1,224,317 hospital episodes in the requested oncology extract during the study period; 1,206,045 episodes (98.5%) had malignant cancer recorded as the principal diagnosis. After identifying patients with metastatic/secondary malignant neoplasm coding in secondary diagnosis fields, the analytic cohort included 272,809 patients (Table 1).
Table 1Participant flow and cohort accounting
Step Records/patients
Source hospital episodes in oncology extract 1,224,317 episodes
Patients with secondary malignant neoplasm/metastatic coding C77-C79 272,809 patients
Patients with psychotic-disorder coding F20-F29 in metastatic care window 1989 patients
Patients with F20-F29 before or at first metastatic-coded episode 1938 patients
Patients with both F20-F29 in the window and before/at first metastatic-coded episode 1889 patients
Patients with F20-F29 in the window only after first metastatic-coded episode 100 patients
Patients with F20-F29 before/at first metastatic-coded episode but not in the window 49 patients
Patients retained in C78-C79 restricted sensitivity cohort 205,136 patients
Psychotic-disorder coding was present in 1989 of 272,809 patients (0.73%). A temporal audit identified 1938 patients with F20-F29 coding before or at the first metastatic-coded episode in the observed extract. These temporal categories were not fully nested: 1889 patients had F20-F29 coding both within the metastatic care window and before or at the index episode, 100 had F20-F29 coding in the metastatic care window only after the index episode, and 49 had F20-F29 coding before or at the index episode but not within the metastatic care window. Among patients with F20-F29 coding in the metastatic care window, the diagnostic-group distribution was: F20 1,421; F21 14; F22 221; F23 18; F25 184; F28 2; F29 164; and no patients coded with F24, F26, or F27. Diagnostic groups were not mutually exclusive because individual patients could have codes from more than one F20-F29 category during the metastatic care window.
Patients with psychotic-disorder coding were younger on average and were more often male. Lung cancer was more frequent among patients with psychotic-disorder coding than among those without psychotic-disorder coding.
Observation opportunities were similar by psychotic-disorder coding status. Patients without psychotic-disorder coding had a mean of 1.43 eligible episodes (median 1, IQR 1–2) and mean observed window duration of 68.6 days (median 11, IQR 5–39). Patients with psychotic-disorder coding had a mean of 1.44 eligible episodes (median 1, IQR 1–2) and mean observed window duration of 62.6 days (median 15, IQR 6–41) (Table 2).
Table 2Observation opportunity during the metastatic care window by psychotic-disorder coding status
Psychotic disorder Episodes, mean (SD) Episodes, median (IQR) Observed days, mean (SD) Observed days, median (IQR)
No 1.43 (1.19) 1 (1–2) 68.6 (160.5) 11 (5–39)
Yes 1.44 (0.80) 1 (1–2) 62.6 (148.2) 15 (6–41)
Patient characteristics and hospital outcomes
Table 3 summarizes patient characteristics and hospital outcomes by psychotic-disorder coding status. Tumor groups are summarized in the main table and shown completely in the supplementary material so that categories sum to the analytic cohort total; not-recorded sex is explicitly displayed.
Table 3Patient characteristics and hospital outcomes by psychotic-disorder coding status
Characteristic Total No psychotic disorder Psychotic disorder
Patients, n 272,809 270,820 1989
Age, mean (SD) 67.3 (13.7) 67.4 (13.7) 62.9 (11.2)
Sex: Male 143,810 (52.7%) 142,672 (52.7%) 1138 (57.2%)
Sex: Female 128,997 (47.3%) 128,146 (47.3%) 851 (42.8%)
Sex: Not recorded 2 (0.0%) 2 (0.0%) 0 (0.0%)
Tumor group: Other cancer 67,987 (24.9%) 67,564 (24.9%) 423 (21.3%)
Tumor group: Lung 48,160 (17.7%) 47,590 (17.6%) 570 (28.7%)
Tumor group: Colorectal 40,250 (14.8%) 39,996 (14.8%) 254 (12.8%)
Tumor group: Breast 28,940 (10.6%) 28,736 (10.6%) 204 (10.3%)
Tumor group: Pancreas 17,399 (6.4%) 17,292 (6.4%) 107 (5.4%)
Tumor group: Stomach 11,530 (4.2%) 11,465 (4.2%) 65 (3.3%)
Other tumor groups 58,543 (21.5%) 58,177 (21.5%) 366 (18.4%)
Approximate non-oncologic Charlson-based score, mean (SD) 0.8 (1.1) 0.8 (1.1) 0.8 (1.0)
Eligible hospital episodes in window, mean (SD) 1.4 (1.2) 1.4 (1.2) 1.4 (0.8)
Observed window duration, days, mean (SD) 68.6 (160.4) 68.6 (160.5) 62.6 (148.2)
Hospital palliative-care coding 46,487 (17.0%) 46,061 (17.0%) 426 (21.4%)
ICU admission 25,474 (9.3%) 25,298 (9.3%) 176 (8.8%)
In-hospital chemotherapy coding 15,777 (5.8%) 15,682 (5.8%) 95 (4.8%)
Mechanical ventilation 7100 (2.6%) 7012 (2.6%) 88 (4.4%)
Transfusion 29,066 (10.7%) 28,856 (10.7%) 210 (10.6%)
Tracheostomy 3243 (1.2%) 3228 (1.2%) 15 (0.8%)
Primary outcome
Hospital palliative-care coding was recorded during the metastatic care window in 46,487 patients overall, including 426 patients with psychotic-disorder coding. After adjustment, psychotic-disorder coding was associated with higher odds of hospital palliative-care coding during the metastatic care window (OR 1.35, 95% CI 1.19–1.52, p < 0.001). The marginal adjusted probability of hospital palliative-care coding was 20.6% with psychotic-disorder coding and 17.0% without psychotic-disorder coding, corresponding to an adjusted risk difference of 3.6 percentage points (Table 4).
Table 4Adjusted association between psychotic-disorder coding and hospital outcomes during the metastatic care window
Outcome Events with psychotic disorder Adjusted OR (95% CI) p value Adj. prob., no psychotic disorder Adj. prob., psychotic disorder Adj. risk difference
Hospital palliative-care coding 426 (21.4%) 1.35 (1.19–1.52) < 0.001 17.0% 20.6% + 3.6 pp
ICU admission 176 (8.8%) 1.07 (0.90–1.28) 0.440 9.3% 9.8% + 0.5 pp
In-hospital chemotherapy coding 95 (4.8%) 0.73 (0.59–0.91) 0.005 5.8% 4.5% − 1.3 pp
Mechanical ventilation 88 (4.4%) 1.44 (1.15–1.82) 0.002 2.6% 3.6% + 1.0 pp
Transfusion 210 (10.6%) 1.02 (0.88–1.19) 0.755 10.7% 10.9% + 0.2 pp
Tracheostomy 15 (0.8%) 0.60 (0.32–1.15) 0.127 1.2% 0.9% − 0.3 pp
Secondary outcomes
Psychotic-disorder coding was associated with higher adjusted odds of mechanical ventilation during the metastatic care window (OR 1.44, 95% CI 1.15–1.82, p = 0.002). Intensive care unit admission showed no clear association (OR 1.07, 95% CI 0.90–1.28, p = 0.440).
In-hospital chemotherapy procedure coding was less frequent among patients with psychotic-disorder coding. After adjustment, psychotic-disorder coding was associated with lower odds of in-hospital chemotherapy procedure coding during the metastatic care window (OR 0.73, 95% CI 0.59–0.91, p = 0.005). This outcome should not be interpreted as complete ascertainment of systemic anticancer therapy.
No clear association was observed for transfusion (OR 1.02, 95% CI 0.88–1.19) or tracheostomy (OR 0.60, 95% CI 0.32–1.15) in the primary model.
Sensitivity analyses
Sensitivity analyses supported the main pattern and are reported in the supplementary material. In the C78-C79 restricted cohort, 205,136 patients were included and 1587 had psychotic-disorder coding; hospital palliative-care coding (OR 1.26, 95% CI 1.11–1.43), in-hospital chemotherapy procedure coding (OR 0.72, 95% CI 0.58–0.91), and mechanical ventilation (OR 1.50, 95% CI 1.17–1.93) remained associated in the same directions. In the sensitivity model defining exposure as F20-F29 coding before or at the first metastatic-coded episode, the corresponding estimates were OR 1.28 (95% CI 1.13–1.45), OR 0.67 (95% CI 0.53–0.84), and OR 1.40 (95% CI 1.10–1.77). In a fixed 12-month sensitivity analysis restricted to patients whose first metastatic-coded episode occurred by 31 December 2023, the corresponding estimates were OR 1.36 (95% CI 1.18–1.57), OR 0.71 (95% CI 0.55–0.91), and OR 1.48 (95% CI 1.13–1.94). Results were also similar in models omitting hospital service and in diagnostic sensitivity analyses focused on schizophrenia/schizoaffective coding (F20/F25) or excluding acute/transient psychosis coding (F23). Tracheostomy was specification-sensitive and was not treated as a central finding.
As an additional audit addressing source-extract consistency, excluding patients whose administrative index episode had Z51.5 as the principal diagnosis did not materially change the main estimates for hospital palliative-care coding (OR 1.34, 95% CI 1.19–1.51), in-hospital chemotherapy procedure coding (OR 0.71, 95% CI 0.57–0.89), or mechanical ventilation (OR 1.45, 95% CI 1.15–1.83).
Discussion
In this nationwide hospital administrative oncology extract, psychotic-disorder coding was associated with different hospital care patterns during the metastatic care window. Patients with psychotic-disorder coding had higher adjusted odds of hospital palliative-care coding and mechanical ventilation, and lower adjusted odds of in-hospital chemotherapy procedure coding.
The higher likelihood of hospital palliative-care coding may have several interpretations. It may reflect greater symptom burden, higher psychosocial complexity, more frequent recognition of supportive care needs, later presentation with greater clinical deterioration, or different documentation practices during admissions involving psychiatric comorbidity. Because administrative data capture coding rather than direct care quality, this finding should not be interpreted as definitive evidence of better, earlier, more specialist, or more appropriate palliative care access. However, it indicates that psychotic-disorder coding is associated with measurable differences in hospital pathways among patients with metastatic/secondary malignant neoplasm coding.
Our hospital palliative-care coding finding is supported by two particularly relevant French studies in patients with schizophrenia and cancer. Gravier et al. reported higher access to palliative care among cancer patients with schizophrenia in a national multicenter retrospective cohort. Viprey et al., in a French national population-based study of patients with lung cancer and schizophrenia, also found higher use of palliative care among patients with schizophrenia [23, 24]. These studies are aligned with our main finding, while differences in outcome definition and care setting should be kept in mind.
At the same time, the literature is not uniform. Other studies have reported lower access to specialist palliative care, disparities in end-of-life care, or potentially worse care trajectories among patients with psychosis or severe mental illness [25–28]. Taken together, the evidence suggests that the relationship between psychotic disorders and palliative oncology care is not unidirectional. It likely depends on the health system, the setting studied, the type of palliative care measured, and whether the outcome captures specialist palliative care, hospital coding, community care, or end-of-life intensity.
The mechanical ventilation result is notable because it points toward higher coding of an acute life-sustaining intervention. This may reflect acute clinical complexity, delayed presentation, differences in goals-of-care documentation, or greater difficulty aligning care plans before hospital deterioration. The absence of a statistically clear association with ICU admission suggests that ventilation and ICU use may capture overlapping but not identical care processes in administrative data.
The in-hospital chemotherapy procedure coding finding is clinically important but must be interpreted cautiously. The adjusted estimate suggests lower odds of hospital-coded chemotherapy procedures among patients with psychotic-disorder coding. This result is not definitive evidence of undertreatment and does not capture outpatient or day-hospital systemic anticancer therapy. Potential explanations include performance status, comorbidity, contraindications, patient preferences, adherence challenges, fragmented oncology follow-up, poorer care coordination, diagnostic overshadowing, reduced oncology engagement, stigma, or barriers in shared decision-making. Future studies linking administrative records with clinical staging, performance status, treatment intent, psychiatric history, and outpatient oncology data are needed.
Lower in-hospital chemotherapy procedure coding and higher mechanical ventilation are separate group-level associations measured during the metastatic care window. They do not establish that these outcomes occurred in the same individuals, occurred in a specific temporal order, or that one type of care replaced another. The pattern is therefore best interpreted as hypothesis-generating evidence of different hospital coding and care patterns among patients with psychotic-disorder coding.
The C78-C79 sensitivity analysis reduced concern that the findings were driven solely by inclusion of C77 lymph-node secondary malignant neoplasm coding. Similar numbers of eligible hospital episodes and observed window duration by psychotic-disorder coding status also reduce, but do not eliminate, concern that higher palliative-care coding simply reflected more opportunities for coding.
This study has several strengths. It uses a large nationwide administrative dataset, applies the patient as the unit of analysis, focuses on metastatic/secondary malignant neoplasm coding in an oncology hospital extract, and evaluates both hospital palliative-care coding and selected markers of hospital treatment intensity. The analytic choices were defined to be clinically interpretable, particularly the use of a metastatic care window beginning at the first hospital episode with documented secondary malignant neoplasm/metastatic coding.
The study also has limitations. Administrative data are subject to coding variability and cannot confirm treatment intent, appropriateness, symptom burden, performance status, patient preferences, outpatient treatment, care quality, referral timing, or the content of palliative-care interventions. The source extract was an oncology administrative extract composed overwhelmingly of episodes with cancer as the principal diagnosis; it therefore may omit admissions primarily coded for infection, respiratory failure, psychiatric deterioration, or other non-cancer diagnoses. Hospital palliative-care coding should not be interpreted as direct measurement of specialist access, quality of care, or patient-centeredness. Psychotic disorders may be undercoded, especially if psychiatric diagnoses were not clinically salient during cancer admissions. Although most patients with F20-F29 coding in the metastatic care window had such coding before or at the index episode in the observed extract, administrative data cannot establish clinical onset. Patient linkage relied on the official recoded identifier supplied in the RAE-CMBD extract; the dataset did not include documentation allowing us to independently audit the identifier-generation algorithm or verify linkage stability across hospitals, autonomous communities, and calendar years. The RAE-CMBD extract did not include a hospital identifier or death date, preventing hospital-level cluster-robust inference and formal survival or competing-risk analysis. Chemotherapy administered outside hospital admission may not be captured. Residual confounding is possible.
Conclusion
Among patients with metastatic/secondary malignant neoplasm coding in a Spanish hospital oncology administrative extract, psychotic-disorder coding was associated with higher hospital palliative-care coding and mechanical ventilation, and with lower in-hospital chemotherapy procedure coding. These findings identify distinct hospital coding patterns and support further study of integrated oncology, psychiatry, and palliative-care pathways for patients with advanced cancer and psychotic disorders.
Supplementary Information
Below is the link to the electronic supplementary material.
References
1.Fiorillo A Sartorius N Mortality gap and physical comorbidity of people with severe mental disorders: the public health scandal Ann Gen Psychiatry 2021 20 52 10.1186/s12991-021-00374-y 34903254 Fiorillo A, Sartorius N (2021) Mortality gap and physical comorbidity of people with severe mental disorders: the public health scandal. Ann Gen Psychiatry 20. 10.1186/s12991-021-00374-y PMC8670051 34903254
2.Chan JKN, et al. Life expectancy and years of potential life lost in people with mental disorders: a systematic analysis. eClinicalMedicine. 2023;65:102294. 10.1016/j.eclinm.2023.102294 PMC10641487 37965432
3.Irwin KE Henderson DC Knight HP Pirl WF Cancer care for individuals with schizophrenia Cancer 2014 120 3 323 334 10.1002/cncr.28431 24151022 Irwin KE, Henderson DC, Knight HP, Pirl WF (2014) Cancer care for individuals with schizophrenia. Cancer 120(3):323–334. 10.1002/cncr.28431 24151022
4.Bentson TM Floe LE Neergaard MA Barriers in cancer trajectories of patients with pre-existing severe mental disorders: a systematic review Psychooncology 2023 32 6 862 874 10.1002/pon.6138 37076979 Bentson TM, Floe LE, Neergaard MA (2023) Barriers in cancer trajectories of patients with pre-existing severe mental disorders: a systematic review. Psychooncology 32(6):862–874. 10.1002/pon.6138 37076979
5.Vendelsoe ANH Stie M Hjorth P Sondergaard J Hansen DG Jensen LH “I just felt that everything came tumbling down around me”-barriers in cancer care for patients with severe mental illness: a qualitative study PLoS ONE 2025 20 1 e0314313 10.1371/journal.pone.0314313 39869604 Vendelsoe ANH, Stie M, Hjorth P, Sondergaard J, Hansen DG, Jensen LH (2025) “I just felt that everything came tumbling down around me”-barriers in cancer care for patients with severe mental illness: a qualitative study. PLoS ONE 20(1):e0314313. 10.1371/journal.pone.0314313 PMC11771855 39869604
6.Woods A Willison K Kington C Gavin A Palliative care for people with severe persistent mental illness: a review of the literature Can J Psychiatry 2008 53 11 725 736 10.1177/070674370805301104 19087466 Woods A, Willison K, Kington C, Gavin A (2008) Palliative care for people with severe persistent mental illness: a review of the literature. Can J Psychiatry 53(11):725–736. 10.1177/070674370805301104 19087466
7.Donald E Stajduhar K A scoping review of palliative care for persons with severe persistent mental illness Palliat Support Care 2019 17 4 479 487 10.1017/S1478951519000087 30887934 Donald E, Stajduhar K (2019) A scoping review of palliative care for persons with severe persistent mental illness. Palliat Support Care 17(4):479–487. 10.1017/S1478951519000087 30887934
8.den Boer K de Veer AJE Schoonmade LJ Verhaegh KJ van Meijel B Francke AL A systematic review of palliative care tools and interventions for people with severe mental illness BMC Psychiatry 2019 19 106 10.1186/s12888-019-2078-7 PMC6446277 30943927 den Boer K, de Veer AJE, Schoonmade LJ, Verhaegh KJ, van Meijel B, Francke AL (2019) A systematic review of palliative care tools and interventions for people with severe mental illness. BMC Psychiatry 19:106. 10.1186/s12888-019-2078-7 PMC6446277 30943927
9.Wilson R Hepgul N Higginson IJ Gao W End-of-life care and place of death in adults with serious mental illness: a systematic review and narrative synthesis Palliat Med 2020 34 1 49 68 10.1177/0269216319867847 31608768 Wilson R, Hepgul N, Higginson IJ, Gao W (2020) End-of-life care and place of death in adults with serious mental illness: a systematic review and narrative synthesis. Palliat Med 34(1):49–68. 10.1177/0269216319867847 31608768
10.Baruth JM Ho JB Mohammad SI Lapid MI End-of-life care in schizophrenia: a systematic review Int Psychogeriatr 2021 33 2 129 147 10.1017/S1041610220000915 32618553 Baruth JM, Ho JB, Mohammad SI, Lapid MI (2021) End-of-life care in schizophrenia: a systematic review. Int Psychogeriatr 33(2):129–147. 10.1017/S1041610220000915 32618553
11.Riley K Hupcey JE Kowalchik K Palliative care in severe and persistent mental illness: a systematic review J Hosp Palliat Nurs 2022 24 3 E88 E93 10.1097/NJH.0000000000000855 35285463 Riley K, Hupcey JE, Kowalchik K (2022) Palliative care in severe and persistent mental illness: a systematic review. J Hosp Palliat Nurs 24(3):E88–E93. 10.1097/NJH.0000000000000855 35285463
12.Denduyver J Detraux J Weydts J De Hert M End-of-life care for people with severe and persistent mental illness and a life-limiting disease: an umbrella review Eur Psychiatry 2025 68 1 e49 10.1192/j.eurpsy.2025.2440 40123415 Denduyver J, Detraux J, Weydts J, De Hert M (2025) End-of-life care for people with severe and persistent mental illness and a life-limiting disease: an umbrella review. Eur Psychiatry 68(1). 10.1192/j.eurpsy.2025.2440 PMC12041735 40123415
13.Cunningham R Sarfati D Stanley J Peterson D Collings S Cancer survival in the context of mental illness: a national cohort study Gen Hosp Psychiatry 2015 37 6 501 506 10.1016/j.genhosppsych.2015.06.003 26160056 Cunningham R, Sarfati D, Stanley J, Peterson D, Collings S (2015) Cancer survival in the context of mental illness: a national cohort study. Gen Hosp Psychiatry 37(6):501–506. 10.1016/j.genhosppsych.2015.06.003 26160056
14.Iglay K Santorelli ML Hirshfield KM Williams JM Rhoads G Lin Y Demissie K Impact of preexisting mental illness on all-cause and breast cancer-specific mortality in elderly patients with breast cancer J Clin Oncol 2017 35 36 4012 4018 10.1200/JCO.2017.73.4947 28934000 Iglay K, Santorelli ML, Hirshfield KM, Williams JM, Rhoads G, Lin Y, Demissie K (2017) Impact of preexisting mental illness on all-cause and breast cancer-specific mortality in elderly patients with breast cancer. J Clin Oncol 35(36):4012–4018. 10.1200/JCO.2017.73.4947 28934000
15.Toender A Munk-Olsen T Vestergaard M Larsen JT Suppli NP Dalton SO Impact of severe mental illness on cancer stage at diagnosis and subsequent mortality: a population-based register study Schizophr Res 2018 201 62 69 10.1016/j.schres.2018.05.011 29891274 Toender A, Munk-Olsen T, Vestergaard M, Larsen JT, Suppli NP, Dalton SO et al (2018) Impact of severe mental illness on cancer stage at diagnosis and subsequent mortality: a population-based register study. Schizophr Res 201:62–69. 10.1016/j.schres.2018.05.011 29891274
16.Kisely S, Crowe E, Lawrence D. Cancer-related mortality in people with mental illness. JAMA Oncol. 2020;6(10):1574–1583. https://pubmed.ncbi.nlm.nih.gov/32618553/10.1001/jamapsychiatry.2013.278 23247556
17.Launders N Scolamiero L Osborn DPJ Hayes JF Cancer rates and mortality in people with severe mental illness: further evidence of lack of parity Schizophr Res 2022 246 260 267 10.1016/j.schres.2022.07.008 35858503 Launders N, Scolamiero L, Osborn DPJ, Hayes JF (2022) Cancer rates and mortality in people with severe mental illness: further evidence of lack of parity. Schizophr Res 246:260–267. 10.1016/j.schres.2022.07.008 35858503
18.Kisely S Alotiby M Khaled N Protani MM Soole R Arnautovska U Siskind D Breast cancer treatment disparities in patients with severe mental illness: a systematic review and meta-analysis Psychooncology 2023 32 5 651 662 10.1002/pon.6120 36892099 Kisely S, Alotiby M, Khaled N, Protani MM, Soole R, Arnautovska U, Siskind D (2023) Breast cancer treatment disparities in patients with severe mental illness: a systematic review and meta-analysis. Psychooncology 32(5):651–662. 10.1002/pon.6120 36892099
19.Protani MM Alotiby M Seth R Lawrence D Jordan SJ Logan H Colorectal cancer treatment in people with severe mental illness: a systematic review and meta-analysis Epidemiol Psychiatr Sci 2022 31 e82 10.1017/S2045796022000634 36384819 Protani MM, Alotiby M, Seth R, Lawrence D, Jordan SJ, Logan H et al (2022) Colorectal cancer treatment in people with severe mental illness: a systematic review and meta-analysis. Epidemiol Psychiatr Sci 31. 10.1017/S2045796022000634 PMC9706308 36384819
20.von Elm E Altman DG Egger M Pocock SJ Gotzsche PC Vandenbroucke JP The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies PLoS Med 2007 4 10 e296 10.1371/journal.pmed.0040296 PMC2020495 17941714 von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP (2007) The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med 4(10):e296. 10.1371/journal.pmed.0040296 PMC2020495 17941714
21.Charlson ME Pompei P Ales KL MacKenzie CR A new method of classifying prognostic comorbidity in longitudinal studies: development and validation J Chronic Dis 1987 40 5 373 383 10.1016/0021-9681(87)90171-8 3558716 Charlson ME, Pompei P, Ales KL, MacKenzie CR (1987) A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 40(5):373–383. 10.1016/0021-9681(87)90171-8 3558716
22.Quan H Sundararajan V Halfon P Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data Med Care 2005 43 11 1130 1139 10.1097/01.mlr.0000182534.19832.83 16224307 Quan H, Sundararajan V, Halfon P et al (2005) Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 43(11):1130–1139. 10.1097/01.mlr.0000182534.19832.83 16224307
23.Gravier M, Haffen E, Charlier C, et al. Access to palliative care for cancer patients with schizophrenia in France: a national multicenter retrospective cohort study. Psycho-Oncology. 2020;29(1):151–158. https://pubmed.ncbi.nlm.nih.gov/31677777/
24.Viprey M Pauly V Salas S Baumstarck K Orleans V Llorca PM Palliative and high-intensity end-of-life care in schizophrenia patients with lung cancer: results from a French national population-based study Eur Arch Psychiatry Clin Neurosci 2021 271 8 1577 1587 10.1007/s00406-020-01182-8 32876751 Viprey M, Pauly V, Salas S, Baumstarck K, Orleans V, Llorca PM et al (2021) Palliative and high-intensity end-of-life care in schizophrenia patients with lung cancer: results from a French national population-based study. Eur Arch Psychiatry Clin Neurosci 271(8):1577–1587. 10.1007/s00406-020-01182-8 32876751
25.Butler H O’Brien A Access to specialist palliative care services by people with severe and persistent mental illness: a retrospective cohort study Int J Ment Health Nurs 2018 27 2 737 746 10.1111/inm.12360 28692186 Butler H, O’Brien A (2018) Access to specialist palliative care services by people with severe and persistent mental illness: a retrospective cohort study. Int J Ment Health Nurs 27(2):737–746. 10.1111/inm.12360 28692186
26.Bergqvist J Hedskog S Hedman C Schultz T Strang P Patients with both cancer and psychosis: to what extent do they receive specialized palliative care? Acta Psychiatr Scand 2024 149 4 313 322 10.1111/acps.13666 38369614 Bergqvist J, Hedskog S, Hedman C, Schultz T, Strang P (2024) Patients with both cancer and psychosis: to what extent do they receive specialized palliative care? Acta Psychiatr Scand 149(4):313–322. 10.1111/acps.13666 38369614
27.Chou FHC, Tsai KY, Su CY, Lee CC. End-of-life care for patients with schizophrenia and cancer: a population-based study. Psycho-Oncology. 2017;26(11):1773–1779. https://pubmed.ncbi.nlm.nih.gov/29066257/
28.Videbech NM Valentin JB Knudsen SV Disparity in health care in end-of-life among patients with lung cancer and pre-existing mental disorders: a nationwide cohort study Palliat Med 2025 39 7 803 813 10.1177/02692163251344133 40515566 Videbech NM, Valentin JB, Knudsen SV et al (2025) Disparity in health care in end-of-life among patients with lung cancer and pre-existing mental disorders: a nationwide cohort study. Palliat Med 39(7):803–813. 10.1177/02692163251344133 40515566
29.Mateen FJ Jatoi A Lineberry TW Aranguren D Creagan ET Croghan GA Do patients with schizophrenia receive state-of-the-art lung cancer therapy? A brief report Psychooncology 2008 17 7 721 725 10.1002/pon.1303 18050362 Mateen FJ, Jatoi A, Lineberry TW, Aranguren D, Creagan ET, Croghan GA et al (2008) Do patients with schizophrenia receive state-of-the-art lung cancer therapy? A brief report. Psychooncology 17(7):721–725. 10.1002/pon.1303 PMC2715919 18050362
Come citare questo studio
Belén Sánchez-Barriguete, Carlos Navarro-García. (2026). Hospital care patterns among patients with psychotic-disorder coding and metastatic/secondary malignant neoplasm coding: A nationwide administrative cohort study in Spain.. Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer. DOI: 10.1007/s00520-026-11440-0
Consulta la pubblicazione originale ↗