eTable 1. Sociodemographic Characteristics at Baseline (2000-2001)
eTable 2. Hazard Ratios for Association Between Bipolar Disorder (2001-2002) and Diagnosis With Selected Conditions (2003-2009)
eTable 3. Standardized Mortality Ratios for Association Between Bipolar Disorder (2001-2002) and Selected Cause-Specific Mortality (2003-2008)
Crump C, Sundquist K, Winkleby MA, Sundquist J. Comorbidities and Mortality in Bipolar DisorderA Swedish National Cohort Study. JAMA Psychiatry. 2013;70(9):931-939. doi:10.1001/jamapsychiatry.2013.1394
Bipolar disorder is associated with premature mortality, but the specific causes and underlying
pathways are unclear.
To examine the physical health effects of bipolar disorder using outpatient and inpatient data
for a national population.
Design, Setting, and Participants
National cohort study of 6 587 036 Swedish adults, including 6618 with bipolar
Main Outcomes and Measures
Physical comorbidities diagnosed in any outpatient or inpatient setting nationwide and mortality
(January 1, 2003, through December 31, 2009).
Women and men with bipolar disorder died 9.0 and 8.5 years earlier on average than the rest of
the population, respectively. All-cause mortality was increased 2-fold among women (adjusted
hazard ratio [aHR], 2.34; 95% CI, 2.16-2.53) and men (aHR, 2.03; 95% CI, 1.85-2.23) with bipolar
disorder, compared with the rest of the population. Patients with bipolar disorder had increased
mortality from cardiovascular disease, diabetes mellitus, chronic obstructive pulmonary disease
(COPD), influenza or pneumonia, unintentional injuries, and suicide for both women and men and
cancer for women only. Suicide risk was 10-fold among women (aHR, 10.37; 95% CI, 7.36-14.60) and
8-fold among men (aHR, 8.09; 95% CI, 5.98-10.95) with bipolar disorder, compared with the rest of
the population. Substance use disorders contributed only modestly to these findings. The association
between bipolar disorder and mortality from chronic diseases (ischemic heart disease, diabetes,
COPD, or cancer) was weaker among persons with a prior diagnosis of these conditions (aHR, 1.40; 95%
CI, 1.26-1.56) than among those without a prior diagnosis (aHR, 2.38; 95% CI, 1.95-2.90;
Pinteraction = .01).
Conclusions and Relevance
In this large national cohort study, patients with bipolar disorder died prematurely from
multiple causes, including cardiovascular disease, diabetes, COPD, influenza or pneumonia,
unintentional injuries, and suicide. However, chronic disease mortality among those with more timely
medical diagnosis approached that of the general population, suggesting that better provision of
primary medical care may effectively reduce premature mortality among persons with bipolar
Bipolar disorder is a severe, chronic mental illness with a reported worldwide prevalence of 1% to 5%.1- 8 It is among the leading causes of disability worldwide9 because of its substantial neuropsychiatric and social effects and medical comorbidities.10,11 In the past 20 years, an increasing body of research has found that bipolar disorder is associated with premature mortality,12 including an almost 2-fold risk of cardiovascular disease mortality13 and 15-fold suicide mortality14 compared with the general population. However, the effects of other comorbidities and the underlying pathways remain unclear. These issues have been understudied for bipolar disorder compared with other disorders, such as schizophrenia and unipolar depression.15 The few previous studies have also had important limitations, including an overreliance on inpatient data or community-based samples. The sole use of inpatient data may potentially result in bias and poor generalizability by including only more severe cases, whereas community-based samples are often further limited by insufficient sample sizes. No studies have examined physical comorbidities and mortality in persons with bipolar disorder using complete outpatient and inpatient data for a national population. Such information would advance our understanding of the underlying causes of premature mortality among bipolar disorder patients and help facilitate better strategies to improve the health of this vulnerable population. We conducted a national cohort study of 6.5 million Swedish adults to examine the association between (1) bipolar disorder and physical comorbidities, (2) bipolar disorder and all-cause or cause-specific mortality, and (3) commonly used medications in bipolar disorder and mortality.
The study population consisted of all adults 20 years or older who had lived in Sweden for at least 2 years as of January 1, 2003 (N = 6 587 036). The primary predictor was bipolar disorder (code F31 in the International Statistical Classification of Diseases, 10th Revision [ICD-10]), which was identified by any outpatient or inpatient diagnosis during the preceding 2 years (January 1, 2001, through December 31, 2002) using the Swedish Outpatient Registry and the Swedish Hospital Registry. The Swedish Outpatient Registry contains all primary and secondary outpatient diagnoses nationwide starting in 2001, and the Swedish Hospital Registry contains all primary and secondary hospital discharge diagnoses with nationwide coverage since 1987. These registries are more than 99% complete for this population.16,17 Bipolar subtype information was unavailable.
A secondary predictor of interest was specific medications that are commonly used for maintenance treatment of bipolar disorder (eg, lithium, olanzapine, valproic acid, lamotrigine, quetiapine, risperidone, carbamazepine, and aripiprazole). These medications were identified from all prescriptions nationwide using the Swedish Pharmacy Registry from its inception in July 1, 2005, through December 31, 2009 (4.5 years). This registry contains a record of all prescriptions dispensed by any outpatient or inpatient pharmacy in Sweden, with medications classified according to the Anatomical Therapeutic Chemical system.18,19 This study was approved by the Regional Ethics Committee of Lund University in Sweden.
The study population was followed up for physical comorbidities and mortality for 7 years, from January 1, 2003, through December 31, 2009. Study outcomes included the following specific comorbidities, which were identified by any primary or secondary diagnosis in the Swedish Outpatient Registry or the Swedish Hospital Registry and classified according to ICD-10 codes: cardiovascular disease (ICD-10 codes I00-I99), which was subclassified as hypertension (ICD-10 code I10), ischemic heart disease (IHD) (ICD-10 codes I20-I25), and stroke (ICD-10 codes I60-I66); any cancer (ICD-10 codes C00-C97); diabetes mellitus (ICD-10 codes E10-E14); lipid disorders (ICD-10 code E78); influenza or pneumonia (ICD-10 codes J09-J18); and chronic obstructive pulmonary disease (COPD) (ICD-10 codes J41-J44). Liver disease (ICD-10 codes K70-K77) and kidney disease (ICD-10 codes N00-N29) were also examined but did not have sufficient diagnoses or deaths among bipolar disorder patients to obtain meaningful risk estimates.
We also examined all-cause mortality during the same follow-up period (7 years) and cause-specific mortality for the first 6 years (January 1, 2003, through December 31, 2008; this was the latest date that cause-specific data were available at the time of analysis). Deaths were identified using the Swedish Death Registry, which contains a record of all deaths in Sweden with compulsory reporting nationwide. Cause of death was based on the primary cause and classified according to ICD-10 codes.
Sociodemographic characteristics that may be associated with bipolar disorder and with comorbidities or mortality were identified using national census data from 2000-2001 and were linked to the registry data using an anonymous personal identification number.20 The following were used as adjustment variables: age (modeled simultaneously as continuous and categorical variables), marital status (married or cohabiting, never married, divorced or widowed, or unknown), educational level (compulsory high school or less [≤9 years], practical [ie, vocational] high school or some theoretical [ie, college preparatory] high school [10-11 years], theoretical high school and/or college [≥12 years], or unknown), employment status (employed or nonemployed; nonemployed includes students and homemakers), and income (categorical variable in quartiles or unknown).
In a separate model, we also examined the potential mediating effect of substance use disorders by further adjusting for any outpatient or inpatient diagnosis of alcohol use disorder (ICD-10 code F10) or other substance use disorders (ICD-10 codes F11-F19) during the study period. Other substance use disorders included mental or behavioral disorders due to opioids, cannabinoids, sedatives or hypnotics, cocaine, other stimulants, hallucinogens, tobacco, volatile solvents, or other psychoactive substances.
The Kruskal-Wallis nonparametric test was used to assess for differences in sociodemographic factors or number of health care system contacts comparing bipolar disorder patients with the rest of the population. Cox proportional hazards regression was used to estimate hazard ratios (HRs) and 95% CIs for the association between bipolar disorder and specific comorbidities, all-cause mortality, and cause-specific mortality, compared with the rest of the population (ie, persons without bipolar disorder). These analyses were stratified by sex because of differences in risk estimates. In the comorbidity analyses, individuals were followed up to the earliest outpatient or inpatient diagnosis of the respective comorbidity. In all analyses, individuals were censored at the time of emigration (n = 101 036; 1.5%), as determined by the absence of a Swedish residential address in census data. We used 3 different adjusted models. The first was adjusted for age, the second for age and other sociodemographic factors (marital status, education, employment status, and income), and the third for age, the same other sociodemographic factors, and alcohol and other substance use disorders. In addition, Cox proportional hazards regression was used to estimate HRs and 95% CIs for the association between specific medications (sole use or any use) and all-cause mortality among bipolar disorder patients, using lithium (the most commonly used maintenance medication) as the reference category. The Cox model proportional hazards assumption was evaluated by graphic assessment of log-log plots21 and was met in each of the models. We assessed for first-order interactions between bipolar disorder and sociodemographic factors or substance use disorders for all-cause mortality by repeating the main analyses after stratifying by these factors and using likelihood ratio tests to formally test for interaction. We also assessed for interaction between bipolar disorder and a prior diagnosis of chronic diseases (IHD, diabetes, COPD, or cancer) for mortality from these conditions by computing HRs after stratifying by a prior diagnosis of these conditions and using a likelihood ratio test to formally test for interaction. All statistical tests were 2-sided and used an α level of .05. All analyses were conducted using STATA statistical software (release 11.0; StataCorp LP).22
Because individual smoking data were unavailable or incomplete, we assessed the sensitivity of the results to the potential mediating effect of smoking using previously reported smoking prevalences among persons with bipolar disorder23 and in the general population.24 In this sensitivity analysis, we calculated standardized mortality ratios and 95% CIs for selected smoking-related causes of death (IHD, stroke, influenza or pneumonia, and COPD) as the ratio of observed to expected numbers of cases, assuming a Poisson distribution. The expected numbers of cases were calculated based on mortality rates for the general Swedish population during this study period, standardized by age and sex, using the following formula to adjust for smoking25,26:Smoking-Adjusted Expected Number = ([a × RR + b]/[c × RR + d]) × Expected Numberwhere a and b, respectively, denote the prevalence of smoking and nonsmoking among persons with bipolar disorder; c and d, respectively, denote the corresponding prevalences in the general population; and RR denotes the relative risk of the respective outcome in smokers compared with nonsmokers. We used 70% as the estimated smoking prevalence among persons with bipolar disorder23 and 25% as the estimated smoking prevalence in the general population.24 We used an RR of 2 for IHD, stroke, or influenza or pneumonia mortality and an RR of 15 for COPD mortality among smokers compared with nonsmokers.27,28
In this population of 6 587 036 Swedish adults, 3918 women (0.12% of all women) and 2700 men (0.08% of all men) were diagnosed as having bipolar disorder in any outpatient or inpatient setting during 2001-2002. More than 60% of bipolar disorder patients were 40 to 69 years old, 59.2% were women, and 30.5% were employed (vs 59.7% in the general population) (eTable 1 in the Supplement). They were more likely than the rest of the population to be diagnosed as having alcohol use disorder (12.2% vs 1.8%) or other substance use disorders (9.8% vs 1.1%), to be unmarried, or to have a lower educational level or income (P < .001 for each comparison by Kruskal-Wallis test).
Persons with bipolar disorder had 3 times as many outpatient clinic visits per year than the rest of the population (mean [SD], 3.0 [5.1]; median, 2.1; vs mean [SD], 1.0 [2.7]; median, 0.4; P < .001 by Kruskal-Wallis test) and more than twice as many hospital admissions per year (mean [SD], 1.0 [3.0]; median, 0.4; vs mean [SD], 0.3 [2.4]; median, 0; P < .001 by Kruskal-Wallis test). Information specifying whether outpatient visits or inpatient admissions were psychiatric or nonpsychiatric was unavailable. After adjusting for age and other sociodemographic factors, bipolar disorder patients had an increased risk of diagnosis with influenza or pneumonia (2.4-fold among women and 1.9-fold among men), COPD (2.1-fold among women and 1.7-fold among men), diabetes (1.7-fold among women and 1.6-fold among men), cardiovascular disease (1.3-fold among women and 1.2-fold among men), and specifically stroke (1.6-fold among women and 1.3-fold among men) (eTable 2 in the Supplement).
In contrast, bipolar disorder patients had no increased risk of diagnosis with IHD, hypertension, lipid disorders, or cancer (eTable 2 in the Supplement). Additional adjustment for substance use disorders resulted in modest attenuation of risk estimates, and the association between bipolar disorder and either stroke or COPD diagnosis among men was no longer statistically significant (eTable 2 in the Supplement).
A total of 623 614 deaths (9.5%) occurred in this study population in 43.5 million person-years of follow-up during 2003-2009. Crude mortality rates (per 1000 person-years) among bipolar disorder patients were 28.5 for women (based on 706 deaths) and 30.3 for men (based on 513 deaths) compared with 14.5 for women and 14.1 for men without bipolar disorder. Among women and men with bipolar disorder, respectively, natural causes accounted for 88.3% and 82.4% of all deaths (vs 96.4% and 93.8% in the general population), and suicide accounted for 5.4% and 9.6% of all deaths (vs 0.6% and 1.7% in the general population) (Table 1).
On average, women with bipolar disorder died 9.0 years earlier than other women (mean age, 73.4 vs 82.4 years), and men with bipolar disorder died 8.5 years earlier than other men (mean age, 68.9 vs 77.4 years) (not shown in the table). This life expectancy difference was not fully explained by unnatural deaths. Among all people who died of natural causes, women with bipolar disorder died 7.5 years earlier than other women (mean age, 75.1 vs 82.6 years), and men with bipolar disorder died 6.6 years earlier than other men (mean age, 71.6 vs 78.2 years).
After adjusting for age and other sociodemographic factors, bipolar disorder was strongly associated with increased all-cause mortality (2.3-fold among women and 2.0-fold among men). Both women and men with bipolar disorder had an increased risk of death from IHD, diabetes, COPD, and influenza or pneumonia (Table 1). Among these causes, the largest HRs were for mortality from influenza or pneumonia (3.7-fold among women and 4.4-fold among men), diabetes (3.6-fold among women and 2.6-fold among men), and COPD (2.9-fold among women and 2.6-fold among men). In addition, women (but not men) with bipolar disorder had significantly increased mortality from stroke (2.6-fold), cancer (1.4-fold), and specifically colon cancer (2.1-fold; based on 13 deaths). Further adjustment for substance use disorders resulted in modest attenuation of most risk estimates, but their statistical significance was unchanged (Table 1).
Among people who died of chronic diseases (IHD, diabetes, COPD, or cancer), the proportion with a prior diagnosis of these conditions was similar among bipolar disorder patients vs others (proportion diagnosed >30 days before death: IHD: 26.0% vs 32.2%; P = .07; diabetes: 73.7% vs 69.7%, P = .59; COPD: 69.6% vs 66.5%; P = .66; cancer: 64.0% vs 65.0%, P = .78). Among bipolar disorder patients who died of these conditions, men were slightly less likely than women to have been previously diagnosed as having them (44.6% vs 54.0%; P = .04), and no other sociodemographic differences were found. After adjusting for age and other sociodemographic factors, the association between bipolar disorder and mortality from any of these conditions was weaker among people with a prior diagnosis of them (adjusted HR [aHR], 1.40; 95% CI, 1.26-1.56) than among those without a prior diagnosis (aHR, 2.38; 95% CI, 1.95-2.90; Pinteraction = .01 by likelihood ratio test). Further adjustment for substance use disorders had a negligible effect on these findings (data not shown).
Bipolar disorder also was strongly associated with increased mortality from suicide and unintentional injuries (Table 1). After adjusting for age and other sociodemographic factors, the risk of death from suicide was 10-fold among women and 8-fold among men with bipolar disorder compared with other women or men. Further adjustment for substance use disorders resulted in smaller risk estimates, but they remained highly significant (Table 1).
In analyses of potential interactions, the association between bipolar disorder and all-cause mortality was stronger among women (Pinteraction = .007) and among persons without alcohol use (Pinteraction< .001) or other substance use disorders (Pinteraction< .001). An interaction with age was also found (Pinteraction< .001) and the stratified risk estimates appeared bimodal, with stronger associations at ages 20 to 39 and 70 to 79 years. Further adjustment for cardiovascular disease had little effect on these results and did not adequately explain the stronger association among older adults (data not shown). In addition, a significant interaction was found with educational level (Pinteraction = .01), although after excluding persons with unknown educational level (12.5% of the cohort) this was no longer significant (Pinteraction = .59) (Table 2). No interactions were found with marital status (Pinteraction = .24), employment status (Pinteraction = .26), or income (Pinteraction = .86) (not shown in the table).
A sensitivity analysis to assess the potential mediating effect of smoking revealed that additional adjustment for smoking resulted in a 15% reduction in most risk estimates for IHD, stroke, influenza or pneumonia, or COPD mortality. However, the statistical significance of all risk estimates remained unchanged (see eTable 3 in the Supplement for complete results).
The association between specific medications and mortality was examined among persons who received any outpatient or inpatient diagnosis of bipolar disorder during 2001 to 2009 (n = 32 764), using sole use of lithium as the reference category. Small numbers of deaths limited the precision of risk estimates for specific medications. After adjusting for sociodemographic factors and substance use disorders, any use of aripiprazole, quetiapine, or lamotrigine was associated with significantly lower all-cause mortality (aHRs, 0.7-0.8) compared with those who only used lithium (Table 3). In contrast, sole use of olanzapine and sole or any use of valproic acid, risperidone, or carbamazepine was associated with modestly increased mortality (aHRs, 1.2-1.4) compared with those who only used lithium. Sole use (but not any use) of atypical antipsychotics (aripiprazole, quetiapine, olanzapine, and/or risperidone) was associated with modestly increased mortality that was borderline significant compared with the sole use of lithium. Bipolar disorder patients who used none of the medications in Table 3 had even higher (1.6-fold) all-cause mortality and increased suicide mortality (aHR, 2.06; 95% CI, 1.11-3.83; based on 47 suicides; not shown in the table). Specific medications that also were associated with increased suicide mortality included any use of olanzapine (aHR, 1.89; 95% CI, 1.07-3.32; based on 14 suicides) or valproic acid (aHR, 1.93; 95% CI, 1.06-3.50; based on 9 suicides) compared with those who only used lithium (not shown in the table). Further adjustment for cardiovascular disease and diabetes resulted in slight attenuation of all risk estimates, and a few (eg, for olanzapine and valproic acid) were no longer significant (Table 3). We also examined whether atypical antipsychotics were associated with an increased risk of lipid disorders among bipolar disorder patients, but no associations were found (any use of atypicals: aHR, 1.05; 95% CI, 0.91-1.21; sole use of atypicals: aHR, 1.10; 95% CI, 0.84-1.44).
In this large national cohort study, women and men with bipolar disorder had 2.3-fold and 2.0-fold increased mortality and died 9.0 and 8.5 years prematurely, respectively, compared with the rest of the population. We identified multiple causes, including increased mortality from cardiovascular disease, diabetes, COPD, influenza or pneumonia, unintentional injuries, and suicide among women and men with bipolar disorder and cancer among women with bipolar disorder. Although the highest HRs were for suicide, the leading causes of death were cardiovascular disease and cancer, as in the general population. Substance use disorders explained only a modest part of these findings. Other lifestyle factors, treatment effects, and/or psychiatric comorbidities may also be major contributors and require further elucidation in future studies.
We also found that the association between bipolar disorder and mortality from chronic diseases (ischemic heart disease, diabetes, COPD, or cancer) was much weaker among persons with an earlier diagnosis of these conditions, suggesting that timely medical diagnosis and treatment may effectively reduce mortality among bipolar disorder patients to approach that of the general population. More complete provision of primary, preventive medical care among bipolar disorder patients is needed to reduce early mortality in this vulnerable population.
The 2-fold all-cause mortality that we found among persons with bipolar disorder is in the midrange of previously reported estimates based on hospital discharge data.29- 31 Our cause-specific mortality results are broadly consistent with previously reported 1.5- to 2.5-fold risks for cardiovascular disease,13,30,31 3-fold risks for diabetes,32,33 1.5-fold risk for COPD,34 3- to 4-fold risks for unintentional injuries,30,31 and 15-fold risks for suicide.14 The high suicide risks underscore the importance of suicide risk assessment and intervention strategies in the care of bipolar disorder patients.35,36 We also found that bipolar disorder was strongly associated with increased diagnosis of and mortality from influenza or pneumonia. This finding was consistent with a 3-fold risk of respiratory disease mortality previously reported among bipolar disorder inpatients,31 although, to our knowledge, a specific association with influenza or pneumonia had not been previously examined. In addition, women (but not men) with bipolar disorder had modestly increased mortality from any cancer and specifically a 2-fold mortality from colon cancer. The few previous studies of bipolar disorder and cancer have yielded discrepant results. Our findings are consistent with an earlier Swedish study31 based on hospital discharge data from 1973 to 1995 that reported modestly increased cancer mortality among women but not men with bipolar disorder, whereas other studies have reported modestly increased cancer incidence among men but not women,37 increased cancer incidence among men and women,38 or null results.30,39 These discrepancies might be related to differences in sample selection, sample size, study location, or study period, which varied widely. Bipolar disorder patients in the current study had a somewhat older age distribution and were more predominantly women compared with several previous studies.
In the current study, bipolar disorder was more strongly associated with mortality among women and among persons without substance use disorders. Some30,31 but not all29 previous studies have reported the same sex difference. The reason for such a difference is unclear, and noncausal explanations are possible. Bipolar disorder may have a smaller proportional effect on mortality among men because of their higher baseline mortality rates compared with women. This explanation may also potentially account for larger HRs in the absence of substance use disorders, which are themselves strong independent risk factors for increased mortality. Several previous studies29,30,40,41 have reported effect modification by age, with a stronger association between bipolar disorder and mortality among young adults, whereas a bimodal pattern appeared in this cohort, with stronger associations at ages of 20 to 39 and 70 to 79 years. Specific effect modifiers and the underlying mechanisms are still unclear and warrant additional study using longitudinal sociodemographic and clinical data.
This study adds to the increasing knowledge of contributors to premature mortality from natural causes among bipolar disorder patients. There are multiple underlying mechanisms. Persons with bipolar disorder may be less likely than the general population to receive primary, preventive medical care.42 They also have a higher prevalence of unhealthy lifestyle factors, including smoking,23 other substance misuse,43,44 and obesity.45 Several pathophysiologic mechanisms have been hypothesized to contribute to cardiovascular disease or diabetes in bipolar disorder patients, including proinflammatory cytokines,46 oxidative stress,15,47 hypothalamic-pituitary-adrenal axis dysfunction,48,49 or common genetic factors.32 Metabolic syndrome is common, and adverse treatment effects may be important contributors.50- 52 Second-generation antipsychotics53 and other psychotropic medications54 may increase the risk of cardiovascular disease, diabetes, and mortality via metabolic pathways involving weight gain,55 glucose intolerance, dyslipidemia, and/or cardiac toxicity.56 The current study found evidence of modestly increased mortality among bipolar disorder patients who used carbamazepine, risperidone, or valproic acid or who solely used olanzapine, whereas users of aripiprazole, quetiapine, or lamotrigine had modestly reduced mortality compared with those who solely used lithium. However, those who used none of these medications had even higher all-cause mortality, as well as a 2-fold suicide mortality. This was consistent with previous evidence suggesting that pharmacologic treatment of bipolar disorder may reduce not only suicide risk57 but also premature mortality from natural causes.58,59 These findings should be interpreted with caution because of possible confounding by disease severity or other unmeasured factors and because few patients received monotherapy and precision was limited for specific medications.
To our knowledge, this is the first study to examine the association between bipolar disorder and mortality or comorbidities using complete outpatient and inpatient diagnoses for a national population. Previous studies have relied mainly on hospital discharge data29- 31,34 or case-control data.42 The availability of outpatient diagnoses is an important strength because it allows the inclusion of milder, nonhospitalized comorbidities. This avoids bias that may potentially result from the sole use of inpatient data and enables more generalizable risk estimates. It also allowed us to assess for potential underdiagnosis of important causes of mortality among bipolar disorder patients and effect modification by prior diagnosis of these conditions.
As in other studies of entire national populations, individual data about smoking, exercise, or other direct lifestyle measurements were unavailable. We explored the potential mediating effects of smoking using previously reported smoking rates, but the lack of individual data prevented a more direct assessment, and the potential effects of physical activity and obesity could not be assessed. Other substance use disorders were examined using nationwide outpatient and inpatient diagnoses, but incomplete ascertainment may have resulted in underestimation of their effects. Other psychiatric comorbidities were not examined, and bipolar subtype information was unavailable. The prevalence of bipolar disorder was lower than previously reported prevalences6 and may reflect underascertainment of mild cases. It also is unclear to what extent our findings are generalizable to other health care systems. The substantial health disparities we found between bipolar disorder patients and the rest of the Swedish population may be even larger in other countries without universal health care.
In summary, Swedish adults with bipolar disorder had increased mortality from multiple causes, including cardiovascular disease, diabetes, COPD, influenza or pneumonia, unintentional injuries, and suicide among women or men and cancer among women. Timely medical diagnosis appeared to improve chronic disease mortality among bipolar disorder patients to approach that of the general population. More effective provision of primary, preventive medical care is needed to reduce early mortality among persons with bipolar disorder.
Corresponding Author: Casey Crump, MD, PhD, Department of
Medicine, Stanford University, 211 Quarry Road, Ste 405, MC 5985, Palo Alto, CA 94304
Published Online: July 17, 2013. doi:10.1001/jamapsychiatry.2013.1394.
Author Contributions: Dr J. Sundquist had full
access to all of the data in the study and takes responsibility for the integrity of the data and
the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition of data: K. Sundquist, J. Sundquist.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: Crump.
Critical revision of the manuscript for important intellectual content: All
Statistical analysis: Crump, J. Sundquist.
Obtained funding: J. Sundquist.
Conflict of Interest Disclosures: None
Funding/Support: This study was supported by grant
R01DA030005 from the National Institute of Drug Abuse and
an Avtal om Lakarutbildning och Forskning (Agreement on Medical Training and Research) project
grant, Lund, Sweden.
Role of the Sponsors: The funding agencies had no role in the design and conduct of
the study; in the collection, analysis, and interpretation of the data; or in the preparation,
review, or approval of the manuscript.