Psychiatric disorder as a risk factor for cancer

different analytic strategies produce different findings

Elise Whitley, G. David Batty, Paul A. Mulheran, Catharine R. Gale, David P. Osborn, Per Tynelius, Finn Rasmussen

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

Reported associations between psychiatric disorders and cancer incidence are inconsistent, with cancer rates in psychiatric patients that are variously higher than, similar to, or lower than the general population. Understanding these associations is complicated by difficulties in establishing the timing of onset of psychiatric disorders and cancer, and by the possibility of reverse causality. Some studies have dealt with this problem by excluding patients with cancers predating their psychiatric illness; others have not considered the issue.
We examined associations between psychiatric hospitalization and cancer incidence in a cohort of 1,165,039 Swedish men, and we explored the impact of different analytic strategies on these associations using real and simulated data.
Relative to men without psychiatric hospitalization, we observed consistent increases in smoking-related cancers in those with psychiatric hospitalizations, regardless of analytic approach (eg, hazard ratio = 1.73 [95% confidence interval = 1.52-1.96]). However, associations with cancers unrelated to smoking were highly dependent on analytic strategy. In analyses based on the full cohort, we observed no association or a modest increase in cancer incidence in those with psychiatric hospitalizations (1.14 [1.07-1.22 ]). In contrast, when men whose cancer predated their psychiatric hospitalizations were excluded, future cancer incidence was lower in psychiatric patients (0.72 [0.67-0.78]). Results from simulated data suggest that even modest exclusions of this type can lead to strong artifactual associations.
Psychiatric disorder-cancer incidence associations are complex and influenced by analytic strategy. A better understanding of the temporal relationship between psychiatric disorder and cancer incidence is required.

Original languageEnglish
Pages (from-to)543-550
Number of pages8
JournalEpidemiology
Volume23
Issue number4
DOIs
Publication statusPublished - Jul 2012

Fingerprint

Psychiatry
Neoplasms
Hospitalization
Incidence
Smoking
Causality
Confidence Intervals

Keywords

  • psychiatric disorder
  • cancer
  • psychiatric hospitalization

Cite this

Whitley, Elise ; Batty, G. David ; Mulheran, Paul A. ; Gale, Catharine R. ; Osborn, David P. ; Tynelius, Per ; Rasmussen, Finn. / Psychiatric disorder as a risk factor for cancer : different analytic strategies produce different findings. In: Epidemiology. 2012 ; Vol. 23, No. 4. pp. 543-550.
@article{0b9871ec202d47e0993dd7147adc0cb5,
title = "Psychiatric disorder as a risk factor for cancer: different analytic strategies produce different findings",
abstract = "Reported associations between psychiatric disorders and cancer incidence are inconsistent, with cancer rates in psychiatric patients that are variously higher than, similar to, or lower than the general population. Understanding these associations is complicated by difficulties in establishing the timing of onset of psychiatric disorders and cancer, and by the possibility of reverse causality. Some studies have dealt with this problem by excluding patients with cancers predating their psychiatric illness; others have not considered the issue. We examined associations between psychiatric hospitalization and cancer incidence in a cohort of 1,165,039 Swedish men, and we explored the impact of different analytic strategies on these associations using real and simulated data. Relative to men without psychiatric hospitalization, we observed consistent increases in smoking-related cancers in those with psychiatric hospitalizations, regardless of analytic approach (eg, hazard ratio = 1.73 [95{\%} confidence interval = 1.52-1.96]). However, associations with cancers unrelated to smoking were highly dependent on analytic strategy. In analyses based on the full cohort, we observed no association or a modest increase in cancer incidence in those with psychiatric hospitalizations (1.14 [1.07-1.22 ]). In contrast, when men whose cancer predated their psychiatric hospitalizations were excluded, future cancer incidence was lower in psychiatric patients (0.72 [0.67-0.78]). Results from simulated data suggest that even modest exclusions of this type can lead to strong artifactual associations. Psychiatric disorder-cancer incidence associations are complex and influenced by analytic strategy. A better understanding of the temporal relationship between psychiatric disorder and cancer incidence is required.",
keywords = "psychiatric disorder, cancer, psychiatric hospitalization",
author = "Elise Whitley and Batty, {G. David} and Mulheran, {Paul A.} and Gale, {Catharine R.} and Osborn, {David P.} and Per Tynelius and Finn Rasmussen",
year = "2012",
month = "7",
doi = "10.1097/EDE.0b013e3182547094",
language = "English",
volume = "23",
pages = "543--550",
journal = "Epidemiology",
issn = "1044-3983",
number = "4",

}

Psychiatric disorder as a risk factor for cancer : different analytic strategies produce different findings. / Whitley, Elise; Batty, G. David; Mulheran, Paul A.; Gale, Catharine R.; Osborn, David P.; Tynelius, Per; Rasmussen, Finn.

In: Epidemiology, Vol. 23, No. 4, 07.2012, p. 543-550.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Psychiatric disorder as a risk factor for cancer

T2 - different analytic strategies produce different findings

AU - Whitley, Elise

AU - Batty, G. David

AU - Mulheran, Paul A.

AU - Gale, Catharine R.

AU - Osborn, David P.

AU - Tynelius, Per

AU - Rasmussen, Finn

PY - 2012/7

Y1 - 2012/7

N2 - Reported associations between psychiatric disorders and cancer incidence are inconsistent, with cancer rates in psychiatric patients that are variously higher than, similar to, or lower than the general population. Understanding these associations is complicated by difficulties in establishing the timing of onset of psychiatric disorders and cancer, and by the possibility of reverse causality. Some studies have dealt with this problem by excluding patients with cancers predating their psychiatric illness; others have not considered the issue. We examined associations between psychiatric hospitalization and cancer incidence in a cohort of 1,165,039 Swedish men, and we explored the impact of different analytic strategies on these associations using real and simulated data. Relative to men without psychiatric hospitalization, we observed consistent increases in smoking-related cancers in those with psychiatric hospitalizations, regardless of analytic approach (eg, hazard ratio = 1.73 [95% confidence interval = 1.52-1.96]). However, associations with cancers unrelated to smoking were highly dependent on analytic strategy. In analyses based on the full cohort, we observed no association or a modest increase in cancer incidence in those with psychiatric hospitalizations (1.14 [1.07-1.22 ]). In contrast, when men whose cancer predated their psychiatric hospitalizations were excluded, future cancer incidence was lower in psychiatric patients (0.72 [0.67-0.78]). Results from simulated data suggest that even modest exclusions of this type can lead to strong artifactual associations. Psychiatric disorder-cancer incidence associations are complex and influenced by analytic strategy. A better understanding of the temporal relationship between psychiatric disorder and cancer incidence is required.

AB - Reported associations between psychiatric disorders and cancer incidence are inconsistent, with cancer rates in psychiatric patients that are variously higher than, similar to, or lower than the general population. Understanding these associations is complicated by difficulties in establishing the timing of onset of psychiatric disorders and cancer, and by the possibility of reverse causality. Some studies have dealt with this problem by excluding patients with cancers predating their psychiatric illness; others have not considered the issue. We examined associations between psychiatric hospitalization and cancer incidence in a cohort of 1,165,039 Swedish men, and we explored the impact of different analytic strategies on these associations using real and simulated data. Relative to men without psychiatric hospitalization, we observed consistent increases in smoking-related cancers in those with psychiatric hospitalizations, regardless of analytic approach (eg, hazard ratio = 1.73 [95% confidence interval = 1.52-1.96]). However, associations with cancers unrelated to smoking were highly dependent on analytic strategy. In analyses based on the full cohort, we observed no association or a modest increase in cancer incidence in those with psychiatric hospitalizations (1.14 [1.07-1.22 ]). In contrast, when men whose cancer predated their psychiatric hospitalizations were excluded, future cancer incidence was lower in psychiatric patients (0.72 [0.67-0.78]). Results from simulated data suggest that even modest exclusions of this type can lead to strong artifactual associations. Psychiatric disorder-cancer incidence associations are complex and influenced by analytic strategy. A better understanding of the temporal relationship between psychiatric disorder and cancer incidence is required.

KW - psychiatric disorder

KW - cancer

KW - psychiatric hospitalization

UR - http://www.scopus.com/inward/record.url?scp=84862126411&partnerID=8YFLogxK

U2 - 10.1097/EDE.0b013e3182547094

DO - 10.1097/EDE.0b013e3182547094

M3 - Article

VL - 23

SP - 543

EP - 550

JO - Epidemiology

JF - Epidemiology

SN - 1044-3983

IS - 4

ER -