cancer mortality among psychiatric patients treated in a community-based system of care: a 25-year...

ORIGINAL PAPER

Cancer mortality among psychiatric patients treatedin a community-based system of care: a 25-year case register study

Giovanni Perini • Laura Grigoletti •

Batul Hanife • Annibale Biggeri • Michele Tansella •

Francesco Amaddeo

Received: 28 January 2013 / Accepted: 17 September 2013 / Published online: 5 October 2013

� Springer-Verlag Berlin Heidelberg 2013

Abstract

Purpose Cancer mortality data allow assessing, at the

same time, the risk of developing the disease and the

quality of care provided to patients after the oncologic

diagnosis. This study explores the risk of death caused by a

single tumor site in a psychiatric population treated in a

community-based psychiatric service.

Methods All patients with an ICD-10 psychiatric diag-

nosis, seeking care in 1982–2006 (25 years), were

included. Data were drawn from the South Verona

Psychiatric Case Register (PCR). Mortality and cause of

death were ascertained using different procedures and

sources. Standardized mortality ratios (SMRs) were used

to compare the observed number of deaths with the

expected number using as reference a population in the

Veneto region.

Results Having been admitted to the hospital (SMR =

1.32), having a short interval from registration (1.52),

having a diagnosis of alcoholism (2.03), and being a

middle-aged male (1.83) were factors showing an increased

risk of death from cancer. Increased SMRs were found for

cancer of the oral cavity (22.93), lymphoma, leukemias,

Hodgkin’s lymphoma (8.01), and central nervous system

(CNS) and cranial nerve tumors (4.75). The SMR

decreased for stomach tumors (0.49). Patients with alco-

holism (5.90 for larynx), affective disorders (20.00 for

lymphomas), and personality disorders (28.00 for SNC)

were found to be exposed to a high risk of cancer death in

specific sites.

Conclusions Psychiatric patients showed different pat-

terns of site-specific cancer mortality when compared with

the general population. The 20-fold higher risk of dying

from hematological neoplasms needs further investigation.

Chronic use of phenothiazines could be involved in the

relative protection from stomach and prostate cancer found

in psychiatric patients.

Keywords Mortality � Cancer � Epidemiological

study � Community-based mental health services �Delivery of health care

Introduction

The occurrence of cancer among psychiatric patients has

been an epidemiologic puzzle for almost a century. Some

incidence studies on cancer in psychiatric populations

report a diminished risk of developing lung, prostate,

breast, and uterine cancer for people suffering from

schizophrenia and other severe mental illnesses [1–3].

Lower p53 antiapoptotic activity has been proposed in

schizophrenic patients, responsible for both altered neuro-

development and cancer protection [4]. Moreover, a spe-

cific antitumor activity has been attributed to some

antipsychotic compounds [5]. On the contrary, other stud-

ies showed increased risk of lung cancer in schizophrenia

[6, 7] and no difference or only slightly higher risk in the

incidence of breast cancer between psychiatric patients and

the general population [8].

G. Perini � L. Grigoletti (&) � B. Hanife � M. Tansella �F. Amaddeo

Section of Psychiatry and Section of Clinical Psychology,

Department of Public Health and Community Medicine,

Policlinico GB Rossi, University of Verona, P.le L.A.

Scuro 10, 37134 Verona, Italy

e-mail: [email protected]

A. Biggeri

DISIA Department of Statistics, Informatics and Applicatins

‘‘G. Parenti’’, University of Florence, Florence, Italy

123

Soc Psychiatry Psychiatr Epidemiol (2014) 49:693–701

DOI 10.1007/s00127-013-0765-0

Cancer mortality data allow assessment of the risk of

developing a disease and the quality of care provided.

Some authors report a reduction in cancer mortality in

schizophrenic patients [9], and some others an excess [10].

These contradictions in psychiatric studies are common

because of methodological problems: studies have been

conducted on different populations in terms of psychiatric

diagnosis and patterns of psychiatric care (inpatients or

outpatients), and they have seldom been controlled for risk

factors, such as smoking, physical activity, and diet. Con-

cern has been expressed about the possibility that psychi-

atric patients receive low-quality medical assistance and/or

oncological early diagnosis and prevention [11, 12]. Most

of the studies report mortality data for ‘cancer’ mortality

without specification of the cancer site [13, 14] because of

the difficulty of collecting large samples of patients with

both a psychiatric diagnosis and cancer.

It is well known that alcoholism is associated with an

increased mortality for some specific cancer sites (such as

the liver and upper aerodigestive cancer) [15]. Fewer

studies have focused on mortality for different cancer sites

among psychiatric non-addicted patients [16–18].

The aim of this study is to explore the risk of death

caused by a single tumor site in a psychiatric population

treated in a community-based system of care compared

with the general population.

Method

The catchment area and the South Verona Psychiatric

Case Register

The study was conducted in South Verona (Veneto region,

Italy), an area of about 100,000 inhabitants that includes

part of Verona (a city of about 260,000 inhabitants in the

Veneto region in northeast Italy) and two neighboring

small towns (Castel d’Azzano and Buttapietra). South

Verona is a mainly urban area with a predominance of

service and manufacturing industries. The study was

approved by the Verona local research ethics committee of

the University Hospital of Verona, Italy (Appr. Prot. No.

1098).

The main agency providing psychiatric care is the South

Verona Community Mental Health Service (CMHS), run

by the Section of Psychiatry and Clinical Psychology of the

University of Verona. The service, set up in 1978 and

monitored ever since by the South Verona Psychiatric Case

Register (PCR), provides a wide range of well-integrated

hospital and community services, ensures continuity of

care through the different phases of treatment and across its

various components, and targets the severely mentally ill.

According to the Italian Psychiatric Reform, since 1978,

in-patient care can only be provided in general hospital

psychiatric wards or private psychiatric clinics. The Italian

National Health Service (NHS) provides medical and

psychiatric care, including admissions to some private

clinics, free of charge [19].

Study design and statistical analyses

All adult patients ([18 years old) with an ICD-10 psy-

chiatric diagnosis, seeking care in 1982–2006 (25 years),

were included in this study, at the time of first contact

(registration in the South Verona PCR). The diagnoses of

all new, clinically assessed cases are routinely reviewed by

the case register staff and are periodically discussed by the

psychiatric clinical staff. In South Verona PCR diagnoses

can be updated. In case of diagnosis update, the last

available diagnosis was used in this study, assuming that

diagnostic accuracy improves over time. Data were drawn

from the South Verona PCR. For the purposes of ICD

diagnoses in the present study, all ICD-9-CM codes were

converted into corresponding ICD-10 codes and were

collapsed into seven diagnostic groups: schizophrenia

(ICD-10 codes F20 through F29; F84), affective disorders

(ICD-10 codes F30; F31; F32; F33; F34.1, 34.8, 34.9;

F38.00; F38.10; F39; F41.2; F43.20, 43.21, 43.22), neu-

rotic, stress-related, and somatoform disorders (ICD-10

codes F40; F41.0, 41.1, 41.3, 41.8, 41.9; F42; F44; F45;

F48; F54), disorders of personality and behavior of the

adult (ICD-10 codes F34.0; F52; F60–F69), drug addiction

(ICD-10 codes F11–19.1, 19.2, 19.3; F55), alcohol addic-

tion (ICD-10 codes F10.1, 10.2, 10.3), and other diagnoses

(including organic psychoses, eating disorders, mental

retardation, and dementia).

Mortality and cause of death were ascertained using

different procedures and sources. First, information about

whether the patients were alive was provided by linkage to

the Central Person Registry of the Municipalities of Ve-

rona, Castel d’Azzano, and Buttapietra. For those not alive,

through linkage with the Mortality Registry of the Local

Health District of Verona, causes of death were estab-

lished. Consultation of the Registries of Deaths of the

Municipalities of Verona, Castel d’Azzano, and Buttapietra

and of the Service of Forensic Medicine of the University

of Verona allowed completion of the missing information.

The cause of death was established in 90 % of cases. These

agencies were able to provide vital status and causes of

death also for people still registered in the area of the

municipality who died outside the study area. All data were

censored on the death date, date of emigration from the

study area (19 % of patients), or 31 December 2006,

whichever happened first.

The standardized mortality ratio (SMR) compares the

observed number of deaths for each cause with the

694 Soc Psychiatry Psychiatr Epidemiol (2014) 49:693–701

123

expected number, i.e., the number of death events that

would be expected in the study population if their death

rates were equivalent to the rates of the general population.

An SMR[1 expresses an excess mortality rate in the study

population, while an SMR between 0 and 1 indicates a

relative protection from death from a specific cause in the

sample studied. Person-years were calculated for each

patient and were divided into five calendar-year periods of

5 years each. The expected number of deaths was obtained

and applied to the corresponding sex and age structure

(person-years) for each of the five periods considered, the

age-, gender-, cause-, and period-specific mortality rates of

the general population in the Veneto Region. These specific

mortality rates were obtained from the Regional Cancer

Register (Registro Veneto dei Tumori); 95 % confidence

intervals were calculated for each SMR assuming a Poisson

distribution of observed cancer deaths using Byar’s

approximation [20]. To address the issue of conducting

multiple tests within the same sample, we applied the false

discovery rate (FDR) control method of Benjamini and

Hochberg [11] across all 168 comparisons using a 10 %

FDR level. In the tables, those tests that passed the Ben-

jamini-Hochberg criterion for selection were highlighted in

bold.

Results

Of the 9,931 patients (4,284 male; 5,647 female; mean age

50.3 ± 17.4) who were followed up from 1982 to 2006, for

a total of 100,762 person-years, 1,518 (15.3 %) died during

the study period. Of the 1,518 deaths registered, 321

(21.1 %) were due to cancer. Regarding the sociodemo-

graphic characteristics of the sample, 4,369 (44 %) of the

patients were married, 7,269 (73.2 %) were living with a

partner, children, or relatives, and only 655 (6.6 %) were in

institutions. The educational level was predominantly pri-

mary (3,019 patients, 30.4 %) or secondary (3,158 patients,

31.8 %), and almost 41 % (4,071 patients) of the sample

was employed, while 41.7 % (4,141) was composed of

housewives, students, or retirees. About 7 % (695) of the

patients had a diagnosis of schizophrenia, 34 % (3,376) had

affective disorders, 20 % (1,986) neurotic and somatoform

disorders, 6 % (595) personality disorders, 3.5 % (347)

alcoholism, 5 % (496) drug addiction, and 24 % (2,383)

other diagnoses (including organic psychoses, eating dis-

orders, mental retardation, dementia, etc.).

Cancer mortality of the cohort was higher than in the

reference population for the middle-aged groups (SMR

1.83 for the age group 25–44, 95 % CI 1.17–2.87;

SMR = 1.27 for the age group 45–64, 95 % CI 1.05–1.53),

in particular for males in the age group 25–44

(SMR = 2.50, 95 % CI 1.42–4.40).

Higher risk of death for cancer was observed for patients

who had been admitted to the hospital (SMR = 1.32, 95 %

CI 1.07–1.61) and within 5 years from registration

(SMR = 1.52 for the interval 0–5 years, 95 % CI

1.30–1.77). Patients with a diagnosis of alcoholism showed

the highest SMR for cancer (2.03, 95 % CI 1.50–2.76),

while for patients with neurosis and somatoform disorders,

there was lower mortality than in the general population

(0.73, 95 % CI 0.53–0.98) (Table 1).

There were few sites of tumors with an increased SMR

(Table 2). The highest risk of death was observed for oral

cavity tumors, with an SMR of 22.93 (95 % CI

9.54–55.10), followed by lymphomas and leukemias, and

in particular non-Hodgkin lymphomas and leukemias

(SMR = 13.52, 95 % CI 9.61–19.01). Brain, CNS, and

cranial nerve tumors carried an almost fivefold greater risk

of death than expected (4.75, 95 % CI 2.55–8.83), while

for digestive system and, in particular, stomach tumors,

there were decreased risks of death (0.74, 95 % CI

0.60–0.91 and 0.49, 95 % CI 0.28–0.84, respectively).

SMRs for each tumor site and each diagnostic group are

shown in Table 3. Patients with alcoholism showed a

Table 1 SMR for cancer mortality by patient characteristics and time

interval from registration

PY O E SMR 95 % CI

All patients (age

groups)

10,0762 321 291.68 1.10 0.99 1.23

In-patients 19,892 96 72.82 1.32* 1.07 1.60

Out-patients 80,464 225 219.03 1.02 0.90 1.17

Diagnosis

Schizophrenia 7,875 17 20.40 0.83 0.50 1.30

Affective disorders 30,887 137 120.09 1.14 0.96 1.34

Neurosis and

somatoform

disorders

21,335 39 53.57 0.73* 0.53 0.98

Personality

disorders

6,274 10 9.71 1.03 0.53 1.83

Alcoholism 4,246 41 20.16 2.03* 1.48 2.73

Drug addiction 7,844 4 3.31 1.21 0.40 2.87

Other diagnoses 22,031 73 64.44 1.13 0.89 1.42

Interval from registration (years)

0–5 41,344 163 107.18 1.52* 1.30 1.77

5–10 26,622 62 73.57 0.84 0.65 1.07

10–15 17,297 50 53.90 0.92 0.70 1.21

15–20 10,712 30 37.83 0.79 0.55 1.12

20–25 4,787 16 19.19 0.83 0.50 1.32

Statistically significant ratios (P \ 0.05) are highlighted in bold

The 95 % CIs were calculated using Byar’s approximation

PY person years, O observed death, E expected death, CI confidence

interval

* P \ 0.05

Soc Psychiatry Psychiatr Epidemiol (2014) 49:693–701 695

123

significant excess risk of death from cancer of the oral

cavity, lip and pharynx, esophagus, larynx, kidney, liver,

other digestive system, and urinary bladder. Alcoholism

was also related to death from melanoma and other skin

tumors (14.90, 95 % CI 3.07–49.31 and 12.36, 95 % CI

1.13–58.28, respectively). Patients with alcoholism also

had a relatively high risk of death from lower airways

cancers (trachea, bronchus, and lung, SMR = 1.73, 95 %

CI 0.85–3.16). In addition, alcoholism was the diagnosis

with the highest SMR for ill-defined tumor sites (8.36,

95 % CI 2.31–22.23). The risk of dying from CNS and

cranial nerve tumors was particularly high among patients

suffering from affective disorders (7.26, 95 % CI

3.26–16.17) and personality disorders (28.02, 95 % CI

7.00–112.05). Relatively low SMRs were found for breast

cancer in patients with schizophrenia (0.31, 95 % CI

0.03–1.47), affective disorders (0.65, 95 % CI 0.36–1.10),

and neurosis (0.64, 95 % CI 0.25–1.42). Differently from

the aggregated cohort, the group of schizophrenic patients

showed a high SMR for prostate cancer (3.23, 95 % CI

0.64–10.34). High SMRs, generated by relatively high

numbers of observed events, were found for non-Hodgkin

lymphomas and leukemias. This was true for all diagnostic

groups with a clear predominance of patients suffering

from affective disorders (20.73, 95 % CI 13.13–31.54).

Discussion

This study demonstrates a different pattern of site-specific

cancer mortality rates in psychiatric patients compared

with the general population.

People seeking and receiving psychiatric assistance,

especially for affective disorders, show a risk of dying from

lymphohematopoietic cancer up to 20 times higher than the

general population. A previous study conducted in the same

area, focused on the analysis of ‘avoidable’ death risk

among psychiatric patients, reported an excess risk of dying

from Hodgkin’s disease (25.68, 95 % CI 6.42–102.68)

within the age range 5–64 [11]. The finding of a specific

strong correlation of a high risk of dying from proliferative

lymphohematopoietic diseases and the diagnosis of affective

disorder needs further investigation. Such a high risk of

death for lymphomas and leukemias could be explained by a

low adherence or tolerance of psychiatric patients to chemo-

or radiotherapeutic programs. Low patient adherence to self-

administered chemotherapy regimens have already been

demonstrated for other tumors [21]. Another possible

explanation for this correlation is that patients could also

develop psychiatric symptoms as a consequence of cancer

pharmacotherapy: high-dose steroids are known to increase

the risk of developing acute psychotic disorders [22] or

depressive episodes [23]. Besides this hypothesis, the causal

correlation seems weak in both directions: there is neither

evidence, so far, of a specific vulnerability of bone marrow

of depressed/manic patients nor any report of a specific risk

for patients with lymphoma or leukemia of developing

abnormal psychological reactions to bad prognoses. Another

possible explanation for this finding is the relative excess of

psychiatric diagnoses and/or referral of patients with blood

diseases due to particular interest or specialization of pro-

fessionals working in hematology departments. The corre-

lation between lymphoma and psychiatric disorders is

partially confirmed by some studies [17, 24], which report,

among male psychiatric patients, a normal incidence of

lymphomas but a high mortality for this cause. Possible

explanations of this discrepancy could be a diagnostic delay,

a difference in quality of medical assistance, or the lack of

secondary prevention programs for psychiatric populations,

as demonstrated elsewhere for other neoplastic and medical

conditions [25, 26]. This difference in the quality of sec-

ondary care for people with mental disorders may also be

due to the influence of stigma and discrimination [27, 28]. It

is possible that the excess mortality in the first years

Table 2 SMR for cancer mortality from neoplasms by type and

tumor site

O E SMR CI 95 %

Oral cavity tumors 5 0.22 22.93* 8.62 49.82

Digestive system tumors 89 119.76 0.74* 0.60 0.91

Stomach tumors 13 26.41 0.49* 0.28 0.82

Respiratory system tumors 82 71.63 1.14 0.92 1.41

Bones and connective tissue

tumors

2 1.34 1.48 0.30 4.78

Melanoma and skin tumors 7 5.23 1.34 0.60 2.63

Breast tumors in woman 31 41.66 0.74 0.52 1.04

Uterine tumors 3 2.65 1.13 0.31 3.02

Prostate tumors 8 15.38 0.52* 0.25 0.98

Testis and other male genital

system tumors

1 0.00 – – –

Bladder, kidney, and other

urinary system tumors

17 11.68 1.45 0.88 2.28

Brain, CNS, and cranialnerves tumors

10 2.10 4.75* 2.44 8.45

Ill-defined and undefined site

tumors

20 15.04 1.32 0.84 2.01

Lymphohematopoietic tissuetumors

35 4.37 8.01* 5.67 11.01

Non-Hodgkin lymphomasand leukemias

33 2.44 13.52* 9.48 18.75

Statistically significant ratios highlighted in bold represent associa-

tions that passed the Benjamini-Hochberg criterion for selection

required by a 10 % false discovery rate correction for multiple testing


O observed death, E, expected death

* P \ 0.05


123

Table 3 SMRs by tumor site and psychiatric diagnostic group

O E SMR 95 % CI O E SMR 95 % CI

1. Oral cavity, lip. and pharynx 2. Esophagus

Schizophrenia 0 0.01 0 – – 0 0.57 0 – –

Affective disorders 1 0.07 14.28 1.30 66.609 2 3.19 0.62 0.13 2.01

Neurosis 0 0.04 0 – – 0 1.50 0 – –

Personality disorders 0 0.01 0 – – 0 0.28 0 – –

Alcoholism 3 0.04 85.13 20.76 246.39 5 0.81 6.18 2.34 13.53

Drug addiction 0 0.01 0 – – 0 0.12 0 – –

Other diagnosis 1 0.05 21.25 1.81 93.24 4 2.03 1.97 0.66 4.68

3. Stomach 4. Colon rectum

Schizophrenia 0 1.66 0 – – 1 1.86 0.53 0.05 2.51


Neurosis 3 4.45 0.67 0.19 1.80 7 4.71 1.48 0.66 2.92


Alcoholism 1 0.84 1.18 0.11 5.55 2 0.91 2.19 0.44 7.04

Drug addiction 0 0.41 0 – – 0 0.40 0 – –

Other diagnosis 2 8.10 0.24 0.05 0.79 5 8.08 0.61 0.23 1.36

5. Liver 6. Gallbladder and biliary tract

Schizophrenia 3 1.19 2.52 0.70 6.73 0 0.57 0 – –


Neurosis 1 3.04 0.32 0.03 1.53 1 1.38 0.72 0.07 3.38


Alcoholism 3 0.92 3.24 0.90 8.70 1 0.15 6.78 0.60 31.08

Drug addiction 0 0.31 0 – – 0 0.08 0 – –

Other diagnosis 3 5.49 0.54 0.15 1.46 2 2.67 0.74 0.15 2.40

7. Pancreas 8. Other digestive system

Schizophrenia 2 1.67 1.19 0.24 3.84 0 0.37 0 – –

Affective disorders 9 10.05 0.89 0.44 1.63 0 2.20 0 – –

Neurosis 1 4.11 0.24 0.02 1.13 0 0.92 0 – –


Alcoholism 1 0.84 1.18 0.11 5.55 1 0.20 5.06 0.45 23.31

Drug addiction 0 0.25 0 – – 0 0.10 0 – –

Other diagnosis 4 6.68 0.59 0.20 1.42 3 1.78 1.68 0.47 4.50

9. Larynx 10. Trachea, bronchus, lung

Schizophrenia 0 0.27 0 – – 1 4.10 0.24 0.02 1.14


Neurosis 2 0.80 2.51 0.50 8.01 8 11.17 0.71 0.34 1.35

Personality disorders 0 0.12 0 – – 4 1.80 2.22 0.74 5.28

Alcoholism 3 0.51 5.90 1.63 15.69 9 5.20 1.73 0.85 3.16

Drug addiction 0 0.08 0 – – 1 1.21 0.82 0.07 3.85

Other diagnosis 0 1.25 0 – – 14 16.99 0.82 0.47 1.35

11. Pleura 12. Bones and connective tissue

Schizophrenia 0 0.11 0 – – 1 0.09 11.01 1.01 51.80

Affective disorders 2 0.64 3.12 0.62 10.02 0 0.50 0 – –

Neurosis 0 0.25 0 – – 0 0.25 0 – –


Alcoholism 0 0.06 0 – – 1 0.06 16.22 1.51 77.70

Drug addiction 0 0 0 – – 0 0.03 0 – –

Other diagnosis 0 0.43 0 – – 0 0.36 0 – –


123

Table 3 continued


13. Melanoma 14. Other skin

Schizophrenia 0 0.20 0 – – 0 0.14 0 -0.01 17.60

Affective disorders 1 1.05 0.95 0.09 4.44 0 0.90 0 0.00 2.74

Neurosis 0 0.61 0 – – 1 0.38 2.64 0.24 12.27

Personality disorders 0 0.12 0 – – 0 0.04 0 -0.04 61.60

Alcoholism 2 0.13 14.90 3.07 49.31 1 0.08 12.36 1.13 58.28

Drug addiction 1 0.11 9.08 0.82 42.38 0 0.04 0 -0.04 61.60

Other diagnosis 1 0.65 1.53 0.14 7.17 0 0.77 0 0.00 3.20

15. Breast (woman) 16. Uterus

Schizophrenia 1 3.17 0.31 0.03 1.47 0 0.21 0 – –


Neurosis 5 7.70 0.64 0.25 1.42 0 0.50 0 – –

Personality disorders 1 0.96 1.04 0.09 4.86 0 0.05 0 – –

Alcoholism 0 0.52 0 -0 – 0 0.03 0 – –

Drug addiction 0 0.10 0 – – 0 0 – – –

Other diagnosis 12 10.79 1.11 0.61 1.88 0 0.65 0 – –

17. Prostate 18. Testis and other male genital system

Schizophrenia 2 0.62 3.23 0.64 10.34 0 0 – – –

Affective disorders 4 4.65 0.86 0.29 2.05 0 0 – – –

Neurosis 1 2.70 0.36 0.03 1.73 0 0 – – –

Personality disorders 0 0.23 0 – – 0 0 – – –

Alcoholism 0 0.61 0 – – 0 0 – – –

Drug addiction 0 0.90 0 – – 0 0 – – –

Other diagnosis 1 5.66 0.17 0.02 0.82 0 0 – – –

19. Bladder and kidney 20. Eye

Schizophrenia 2 0.67 2.96 0.60 9.57 0 0 – – –

Affective disorders 3 4.26 0.70 0.19 1.88 0 0 – – –

Neurosis 3 2.12 1.41 0.39 3.78 0 0 – – –

Personality disorders 1 0.24 4.18 0.38 19.43 0 0 – – –

Alcoholism 3 0.47 6.38 1.77 17.03 0 0 – – –

Drug addiction 0 0.39 0 – – 0 0 – – –

Other diagnosis 5 3.54 1.41 0.54 3.10 1 0 – – –

21. Brain and cranial nerves 22. Ill defined

Schizophrenia 0 0.14 0 – – 0 1.00 0 – –


Neurosis 0 0.41 0 – – 2 2.35 0.85 0.17 2.73

Personality disorders 2 0.07 28.02 5.70 91.58 0 0.27 0 – –

Alcoholism 0 0.15 0 – – 3 0.36 8.36 2.31 22.23

Drug addiction 0 0.01 0 – – 0 0.10 0 – –

Other diagnosis 2 0.49 4.08 0.81 13.08 7 4.69 1.49 0.67 2.93

23. Hodgkin lymphomas 24. Non-Hodgkin lymphomas and leukemias

Schizophrenia 0 0.13 0 – – 2 0.17 11.72 2.35 37.71


Neurosis 0 0.33 0 – – 2 0.46 4.39 0.87 13.94

Personality disorders 0 0.08 0 – – 2 0.06 33.20 6.65 106.85

Alcoholism 0 0.09 0 – – 2 0.13 15.51 3.07 49.31

Drug addiction 0 0.13 0 – – 1 0.02 49.91 4.53 233.10


123

accounts for people who were already sick at the time of

registration. These people might have died faster, the psy-

chiatric disorder being an obstacle to seeking and receiving

standard cancer prevention, early diagnosis, and care.

Longer follow-ups would allow a clearer discrimination

between the excess mortality due to the low medical assis-

tance prior to the registration and the mortality due to psy-

chiatric disorders themselves. The reported high risk of

developing multisite cancer in patients with alcohol abuse/

addiction was largely predictable [15]. The data of an

increase also in airways (upper and lower) cancer mortality

is less obvious among alcohol abusers, and it can be

explained by the high prevalence of cigarette smoking

among such patients. Of interest is also the finding of a high

risk of death for unspecified cancer diagnoses among alco-

hol-addicted patients, leading to the hypothesis of either

frequent diagnostic delays or lack of accurate medical

reporting. Oropharyngeal cancers were also increased in

people with affective disorders; these findings could be

related to the diagnosis, for example, people diagnosed as

having affective disorders may have also had (undiagnosed)

comorbid alcohol misuse or dependency, which could

account for the elevated SMR. These results stress the

importance of increasing the number and improving the

quality of health care and promotion programs targeted to

high-risk populations.

The reported high death risk for CNS and cranial nerve

tumors in the psychiatric population needs further research:

one would expect high SMRs for such causes in the ‘other

diagnosis’ group, which includes the ‘organic psychoses’.

Unexpectedly, we found significantly high mortality ratios

also among patients with affective and personality disor-

ders. As previously hypothesized for lymphomas and leu-

kemias, this could be due the development of psychiatric

symptoms as a consequence of cancer.

It is now clear that antipsychotic treatment influences

all-cause mortality in psychiatric patients [29]. Interest-

ingly, cancer mortality could be influenced by treatment

with neuroleptics [5], as well. The finding of a relative

protection of psychiatric patients from stomach cancer is of

particular interest, regarding the relationship between

neuroleptics and cancer. In our data set, we found stomach

cancer SMRs lower than 1 only for the whole sample and

not for patients with schizophrenia alone. Although the

SMR for stomach cancer in the schizophrenia group was

not calculable, we observed fewer cases than expected in

the schizophrenia group (1.66 expected vs. none observed).

Our data do not allow differentiating the effect on mortality

of the diagnosis itself from those of the pharmacological

treatment, since we did not collect information about

psychiatric medications. Nevertheless, since Saku and

colleagues [18] report an SMR for stomach cancer of 0.27

(95 % CI 0.06–0.78) among males suffering from schizo-

phrenia, our initial findings seemed worth pursuing. One

possible explanation for this stomach cancer protection is

the chronic use of phenothiazines, compounds known to

have antineoplastic activity [5]: neuroleptics could prevent

malignant degeneration of stomach tissues more effectively

than in other organs because of the higher concentration of

the substance found in the gastric lumen. Tran and col-

leagues [30] also reported a higher SMR for stomach

cancer only among schizophrenic women, whereas the total

sample and males showed no statistical significance for this

cause of death.

The relative protection of the male psychiatric popula-

tion from prostate cancer has already been reported [31]

among people treated with high-dose phenothiazines.

There is evidence that prostate cancer cells are sensitive to

phenothiazine cytotoxic activity [32], but this appears to be

in contrast with our finding of a high, although not sig-

nificant, SMR for prostate cancer for the schizophrenia

group, the one more likely to be prescribed this family of

antipsychotics.

We report neither an excess death risk nor protection from

lung and breast cancer in psychiatric patients. Lung cancer

mortality data are in line with those of other epidemiological

studies [16], but they need to be controlled for smoking

habits, in order to reveal a possible underlying protection of

psychiatric patients from this cause of death. This is a con-

troversial result since numerous studies have reported a

higher prevalence of cigarette smoking in schizophrenic

patients than in the general population along with a higher

nicotine, and probably carcinogen, intake [33].

Regarding breast cancer mortality in our population, we

did not confirm the results reported by other studies [24] of

a slight increase in the risk of death for this cause in

Table 3 continued


Other diagnosis 0 0.46 0 – – 4 0.62 6.40 2.16 15.34

Statistically significant ratios highlighted in bold represent associations that passed the Benjamini-Hochberg criterion for selection required by a

10 % false discovery rate correction for multiple testing


O observed death, E expected death


123

psychiatric female patients. In fact, our population showed

relatively low SMRs for breast cancer among people suf-

fering from schizophrenia, affective disorders, and neuro-

sis. This finding is not easy to interpret since data about

parity of individuals were not available, but it could sug-

gest that, in our area, accessibility to secondary breast

cancer prevention programs is maintained for people with

psychiatric disabilities.

One limitation of this study was the small number of

person-years, which did not allow us to have a sufficient

number of events necessary for the calculation of reliable

SMRs. Another limitation was the lack of information about

patients’ habits such as smoking, physical activity, and diet,

which were not available in our data set. This limited the

possibility to disentangle, between other confounders, the

effects due to lifestyle variables. More person-years, which

account at the same time for both more cases and longer

follow-ups, are needed in future studies. Moreover, we need

studies where psychiatric patients are compared with sub-

jects with particular disabilities or lifestyles. This could be

achieved using a case-control design or using SMRs and

taking into account the mortality of these populations. We

also need to study the incidence of cancer among psychiatric

patients to know whether they have an higher incidence of

cancer or only a higher mortality due to worse quality of care,

or a combination of these two effects.

Another possible limitation is that data on all those

patients who migrated outside Verona were truncated at the

date of migration. As we used person-years to calculate

SMRs and in our area there is no evidence of selective

migration, we can exclude that this limitation biased our

results.

Historical cohorts can be less sensitive than inception

cohorts in retrieving death events retrospectively. This

could explain the relatively low cancer mortality rate that

we found for psychiatric patients overall. Nevertheless, the

death certificates we used for the purposes of the present

study are mandatory in Italy for each person who dies in

the area. Thus, loss of information is more likely to derive

from migration out of the area than from lack of infor-

mation about death events. Another possible explanation

for the relatively low mortality rate overall is the relatively

short follow-up. A short follow-up does not allow the

occurrence of the event ‘cancer death’ because of pre-

mature death of psychiatric patients, thus leading to an

underestimation of the risk of dying from cancer [34]. In

our opinion, this distortion, although possible, is probably

minimal in this case. This sort of underestimation is greater

in ‘‘closed’’ cohorts, when subjects are enrolled at the same

time and then followed up without new entrances in the

cohort. The cohort of the present study, in fact, is open and

dynamic, with patients entering the cohort during a long

calendar period.

Future research should address the relationship between

mortality and pharmacological treatment and quality of

medical care patterns provided to psychiatric patients. The

epidemiological analysis of the relationships between

psychiatric diagnoses and cancer mortality can open new

ways to understand the physiopathology of both psychiatric

and neoplastic diseases. Moreover, studies should be con-

ducted in order to verify whether more ‘‘patient-centered’’

patterns of psychiatric care [35, 36] show a positive cor-

relation with cancer mortality.

Acknowledgments This work was supported by Fondazione Cassa

di Risparmio di Verona Vicenza Belluno e Ancona, Biomedical

Research Projects 2003, grant ‘‘Mortality for neoplasm among psy-

chiatric patients and general population’’.

We thank the South-Verona PCR data managers, Cristina Pighi and

Giuliano Meneghelli, for their help in data collection.

This paper was presented at the IXth International Conference of

the European Network for Mental Health Service Evalua-

tion(ENMESH), Ulm, Germany, 23–25 June 2011.

Conflict of interest None for any of the authors.

References

1. Barak Y, Achiron A, Mandel M, Mirecki I, Aizenberg D (2005)

Reduced cancer incidence among patients with schizophrenia.

Cancer 104:2817–2821

2. Barak Y, Levy T, Achiron A, Aizenberg D (2008) Breast cancer

in women suffering from serious mental illness. Schizophr Res

102:249–253

3. Gulbinat W, Dupont A, Jablensky A, Jensen OM, Marsella A,

Nakane Y, Sartorius N (1992) Cancer incidence of schizophrenic

patients. Results of record linkage studies in three countries. Br J

Psychiatry Suppl (18):75–83

4. Park JK, Lee HJ, Kim JW, Park YH, Lee SS, Chang HI, Song JY,

Yoon DJ, Bahn GH, Shin YH, Kim YJ, Kim SA, Choe BK, Kim

CJ, Chung JH (2004) Differences in p53 gene polymorphisms

between Korean schizophrenia and lung cancer patients. Schiz-

ophr Res 67:71–74

5. Choi JH, Yang YR, Lee SK, Kim SH, Kim YH, Cha JY, Oh SW,

Ha JR, Ryu SH, Suh PG (2008) Potential inhibition of PDK1/Akt

signaling by phenothiazines suppresses cancer cell proliferation

and survival. Ann N Y Acad Sci 1138:393–403

6. Grinshpoon A, Barchana M, Ponizovsky A, Lipshitz I, Nahon D,

Tal O, Weizman A, Levav I (2005) Cancer in schizophrenia: is

the risk higher or lower? Schizophr Res 73:333–341

7. Lichtermann D, Ekelund J, Pukkala E, Tanskanen A, Lonnqvist J

(2001) Incidence of cancer among persons with schizophrenia

and their relatives. Arch Gen Psychiatry 58:573–578

8. Oksbjerg-Dalton S, Munk LT, Mellemkjaer L, Johansen C,

Mortensen PB (2003) Schizophrenia and the risk for breast can-

cer. Schizophr Res 62:89–92

9. Cohen M, Dembling B, Schorling J (2002) The association

between schizophrenia and cancer: a population-based mortality

study. Schizophr Res 57:139–146

10. Black DW, Winokur G (1986) Cancer mortality in psychiatric

patients: the Iowa Record-linkage Study. Int J Psychiatry Med

16:189–197

11. Amaddeo F, Barbui C, Perini G, Biggeri A, Tansella M (2007)

Avoidable mortality of psychiatric patients in an area with a


123

community-based system of mental health care. Acta Psychiatr

Scand 115:320–325

12. Chou FH, Tsai KY, Su CY, Lee CC (2011) The incidence and

relative risk factors for developing cancer among patients with

schizophrenia: a nine-year follow-up study. Schizophr Res

129:97–103

13. Guan NC, Termorshuizen F, Laan W, Smeets HM, Zainal NZ,

Kahn RS, De Wit NJ, Boks MP (2012) Cancer mortality in

patients with psychiatric diagnoses: a higher hazard of cancer

death does not lead to a higher cumulative risk of dying from

cancer. Soc Psychiatry Psychiatr Epidemiol

14. Kisely S, Smith M, Lawrence D, Maaten S (2005) Mortality in

individuals who have had psychiatric treatment: population-based

study in Nova Scotia. Br J Psychiatry 187:552–558

15. Zaridze D, Brennan P, Boreham J, Boroda A, Karpov R, Lazarev

A, Konobeevskaya I, Igitov V, Terechova T, Boffetta P, Peto R

(2009) Alcohol and cause-specific mortality in Russia: a retro-

spective case-control study of 48,557 adult deaths. Lancet

373:2201–2214

16. Goldacre MJ, Kurina LM, Wotton CJ, Yeates D, Seagroat V

(2005) Schizophrenia and cancer: an epidemiological study. Br J

Psychiatry 187:334–338

17. Kisely S, Sadek J, MacKenzie A, Lawrence D, Campbell LA

(2008) Excess cancer mortality in psychiatric patients. Can J

Psychiatry 53:753–761

18. Saku M, Tokudome S, Ikeda M, Kono S, Makimoto K, Uchimura

H, Mukai A, Yoshimura T (1995) Mortality in psychiatric

patients, with a specific focus on cancer mortality associated with

schizophrenia. Int J Epidemiol 24:366–372

19. Amaddeo F, Burti L, Ruggeri M, Tansella M (2009) Long-term

monitoring and evaluation of a new system of community-based

psychiatric care. Integrating research, teaching and practice at the

University of Verona. Ann Ist Super Sanita 45:43–53

20. Breslow NE, Day NE (1987) Statistical methods in cancer

research: the design and analysis of cohort studies, vol 2. Inter-

national Agency for Research on Cancer, Lyon

21. Lebovits AH, Strain JJ, Schleifer SJ, Tanaka JS, Bhardwaj S,

Messe MR (1990) Patient noncompliance with self-administered

chemotherapy. Cancer 65:17–22

22. Cerullo MA (2008) Expect psychiatric side effects from corti-

costeroid use in the elderly. Geriatrics 63:15–18

23. Miller AH, Ancoli-Israel S, Bower JE, Capuron L, Irwin MR

(2008) Neuroendocrine-immune mechanisms of behavioral

comorbidities in patients with cancer. J Clin Oncol 26:971–982

24. Lawrence D, Holman CD, Jablensky AV, Threlfall TJ, Fuller SA

(2000) Excess cancer mortality in Western Australian psychiatric

patients due to higher case fatality rates. Acta Psychiatr Scand

101:382–388

25. Druss BG, Bradford DW, Rosenheck RA, Radford MJ, Krumholz

HM (2000) Mental disorders and use of cardiovascular proce-

dures after myocardial infarction. JAMA 283:506–511

26. O’Rourke RW, Diggs BS, Spight DH, Robinson J, Elder KA,

Andrus J, Thomas CR, Hunter JG, Jobe BA (2008) Psychiatric

illness delays diagnosis of esophageal cancer. Dis Esophagus

21:416–421

27. Lasalvia A, Tansella M (2008) Fighting discrimination and

stigma against people with mental disorders. Epidemiol Psichiatr

Soc 17:1–9

28. Thornicroft G (2008) Stigma and discrimination limit access to

mental health care. Epidemiol Psichiatr Soc 17:14–19

29. Tiihonen J, Lonnqvist J, Wahlbeck K, Klaukka T, Niskanen L,

Tanskanen A, Haukka J (2009) 11-year follow-up of mortality in

patients with schizophrenia: a population-based cohort study

(FIN11 study). Lancet

30. Tran E, Rouillon F, Loze JY, Casadebaig F, Philippe A, Vitry F,

Limosin F (2009) Cancer mortality in patients with schizophre-

nia: an 11-year prospective cohort study. Cancer 115:3555–3562

31. Mortensen PB (1992) Neuroleptic medication and reduced risk of

prostate cancer in schizophrenic patients. Acta Psychiatr Scand

85:390–393

32. Kurihara T, Nojima K, Sakagami H, Motohashi N, Molnar J

(1999) Electronic structure and cytotoxic activity of ‘‘half-mus-

tard type’’ phenothiazines by MM3 and PM3 methods. Antican-

cer Res 19:3895–3899

33. Dalack GW, Healy DJ, Meador-Woodruff JH (1998) Nicotine

dependence in schizophrenia: clinical phenomena and laboratory

findings. Am J Psychiatry 155:1490–1501

34. Bushe CJ, Hodgson R (2010) Schizophrenia and cancer: in 2010

do we understand the connection? Can J Psychiatry 55(12):

761–767

35. Gask L, Coventry P (2012) Person-centred mental health care: the

challenge of implementation. Epidemiol Psychiatr Sci 21:

139–144

36. Salvador-Carulla L, Mezzich JE (2012) Person-centred medicine

and mental health. Epidemiol Psychiatr Sci 21:131–137


123

cancer mortality among psychiatric patients treated in a community-based system of care: a 25-year...

Documents