Specific causes of death in patients with bullous pemphigoid as measured by death certificate data: a retrospective cohort study

Authors


  • Funding: This study was supported by Mayo Clinic Department of Dermatology funds and the Rochester Epidemiology Project (grant no. R01 AG034676 from the National Institute on Aging).
  • Conflicts of interest: None.

Abstract

Background

Mortality rates in patients with bullous pemphigoid (BP) are higher than those in age-matched counterparts. However, the specific causes of death in BP subjects have not been evaluated systematically.

Objectives

We sought to characterize the causes of death in patients with BP as recorded by death certificate and to compare these with death data for age- and location-matched control subjects.

Methods

This was a retrospective cohort analysis conducted in a large tertiary referral center. Twenty-seven participants who had a confirmed antemortem diagnosis of BP, were residents of Olmsted County, Minnesota, and had died between January 1, 1999, and January 1, 2009, were included in the study. Underlying cause of death and multiple causes of death data for the study population were compared with data sourced from the Centers for Disease Control and Prevention (CDC) for a control group matched by age and geographic location of origin by International Classification of Diseases, 10th Revision (ICD-10) block, and specific ICD-10 codes.

Results

Comparisons of specific ICD-10 codes revealed increased rates of sepsis (P = 0.031), dementia (P = 0.049), and major depressive disorder (P = 0.005) in the study group. The collective incidence of ICD-10 codes for infections indicated that infections were more frequent contributors to death in the study group (P = 0.035).

Conclusions

Clinicians should be mindful of contributors to death in patients with BP and might consider screening for mental health issues, educating patients on the early symptoms of sepsis, and minimizing risk factors for infection.

Introduction

Bullous pemphigoid (BP), the most common autoimmune bullous dermatosis,[1] is characterized by a humoral response to the BP230 and BP180 self-antigens. Both antigens are components of the adhesion complex, which secures epidermal basal cells to the underlying basement membrane. The disease manifests as pruritic, tense bullae and urticarial plaques in the elderly population.

Recent studies have noted that the incidence of BP is increasing.[2, 3] Some have also reported an increased rate of mortality in BP subjects compared with the age- and sex-adjusted general population,[2, 4] although this finding has not been reported consistently.[5] In the USA, the 1-year mortality rate for patients diagnosed with BP has been reported to be 23%.[5] Common comorbidities and associated medical conditions have also been studied. Several studies attribute the increased mortality to advanced age, the presence of associated medical conditions, and hospitalization, rather than disease-specific factors.[6, 7] Others suggest that several degenerative neurological diseases and stroke may be associated with BP.[4, 8-10]

Although many studies have addressed the incidence and rate of mortality, we are aware of scant literature that has addressed the specific causes of mortality in patients with BP. The purpose of the present study is to characterize the underlying cause of death (UCD) and multiple causes of death (MCD) amongst patients with BP from Olmsted County, Minnesota, seen at the Mayo Clinic, through a retrospective chart and death certificate review.

Materials and methods

This was a retrospective cohort study which included the review of chart and death certificate data. Because all patients were deceased, this project was deemed not to require institutional review board approval.

Deceased patients with an antemortem diagnosis of BP were included in the study. Inclusion criteria required the patient to have died between January 1, 1999, and January 1, 2009, to have been a resident of Olmsted County, Minnesota, with death certification, and to have been diagnosed with BP based on clinical, histological, and immunological criteria. To ensure accuracy in the diagnosis of pemphigoid, a meticulous chart review was performed to ensure that clinical features of pemphigoid were present, and clinical features of other mimicking diseases were absent.[11] In addition to compatible clinical findings, two of the four following pathological findings were required: (i) skin microscopic findings consistent with BP (eosinophilic subepidermal blister, urticarial tissue reaction and/or eosinophilic spongiosis); (ii) direct immunofluorescence with linear C3+/− immunoglobulin G (IgG); (iii) positive indirect immunofluorescence; and (iv) a positive enzyme-linked immunosorbent assay (ELISA) for BP180 and/or BP230 autoantibodies. Therefore, evidence of tissue or serum circulating autoantibodies directed against the basement membrane zone was available for each patient.

Patients were identified in a search through the Mayo Clinic patient record database. Death was confirmed using the Rochester Epidemiology Project (REP) death browser database. Twenty-seven deceased patients from Olmsted County were found to meet the study criteria and were included in the study. No eligible patients were excluded.

Diagnostic findings of BP and demographic information were recorded from electronic medical charts. Data on UCD and MCD were obtained from the records in the REP death browser database. The US Centers for Disease Control and Prevention (CDC) define UCD as “the disease or injury that initiated the train of events leading directly to death, or the circumstances of the accident or violence, which produced the fatal injury”.[12] Up to 20 additional factors contributing to death (MCD) are also reported on the standard US death certificate. We evaluated both UCD and MCD data because it is known that both should be evaluated together in order to establish the most accurate picture of the contributing causes of death.[13] For standardization, the UCD for each subject was also listed in the MCD category. Each International Classification of Diseases, 10th Revision (ICD-10), code documented in the death certificates was sorted into ICD-10 blocks for data analysis for both UCD and MCD.

Data for UCD and MCD from the CDC mortality database (http://wonder.cdc.gov/) were obtained for a group of subjects matched by age and geographic location in Olmsted County, who died during the period of study, to serve as a control group.[14] Mortality data for patients with BP were compared with CDC data by ICD-10 block and specific ICD-10 codes for both UCD and MCD.

Data for MCD were evaluated similarly to those for UCD with one important difference. Whereas each patient was designated with one UCD code, many MCDs were listed as contributing factors in each death certificate. Therefore, MCD data were compared by proportions of all MCD codes, rather than by proportions of all patients. For example, the death certificates of our 27 BP patients contained 27 ICD-10 codes for UCD and a total of 107 codes for MCD. When our patient group had been subtracted from the control group, the CDC was found to report 6802 UCD and 25,285 MCD codes representing 6802 deaths. The average number of MCD entries per subject was 4.0 in the study population and 3.7 in the control group. Proportions were compared by ICD-10 block and specific ICD-10 codes.

The ICD-10 is structured to organize codes pertaining to infection into the block A00–B99; however, many causes of death with underlying infection are listed by organ system. Therefore, in order to further evaluate infection, all ICD-10 codes with an infectious etiology were compared with all listed MCD mortality with underlying infection. One code in the study group, K83 (cholangitis), was excluded as the CDC did not include data for cholangitis because of its rarity.

Comparisons of causes of death between the study and control groups were evaluated using chi-squared and Fisher's exact tests. Statistical analyses were performed using SAS Version 9.2 (SAS Institute, Inc., Cary, NC, USA). All tests were two-sided. P-values of < 0.05 were considered to indicate statistical significance.

Results

In all, 27 subjects with confirmed BP fulfilled our study criteria. Of these, 22 (81%) were women. The median age at death was 89.6 years, and the mean ± standard deviation (SD) age at death was 88.5 ± 7.6 years (range: 66.8–97.7 years). The median time from diagnosis of BP to death was 2.5 years (mean ± SD: 3.1 ± 2.6 years).

The UCD and MCD were compared between patients with BP and those sourced from CDC data.

The comparison of UCD data revealed a significant difference in the category of neoplasms, with the control population experiencing more neoplasms than the study group (P = 0.043). When specific ICD-10 codes were reviewed, no statistically significant differences were found between the groups (Tables 1 and 2).

Table 1. Underlying cause of death (UCD) as reported by ICD-10 block for subjects in the bullous pemphigoid (BP) group and the Centers for Disease Control (CDC) groupa
 UCD by ICD-10 blockBP group, % n CDC group, % n P-value
  1. a

    No UCD data were recorded in the study or control group in the following ICD-10 blocks: H00–H57 Diseases of the eye and adnexa; H60–H93 Diseases of the ear and mastoid process; O00–O91 Pregnancy, childbirth, puerperium; P00–P96 Certain conditions originating in the perinatal period; Q00–Q99 Congenital malformations, deformations, and chromosomal abnormalities, and U00–U99 Codes for special purposes

  2. ICD-10, International Classification of Diseases, 10th Revision.

A00–B99Certain infections and parasitic diseases3.7011.21820.28
C00–D48Neoplasms7.41224.0816380.043
D50–D89Diseases of the blood and blood-forming organs and certain disorders involving the immune system000.28191.0
E00–E88Endocrine, nutritional and metabolic diseases3.7012.251530.46
F01–F99Mental and behavioral disorders18.5258.675900.08
G00–G98Diseases of the nervous system7.4126.854660.71
I00–I99Diseases of the circulatory system29.63833.7122930.65
J00–J98Diseases of the respiratory system11.11311.297681.0
K00–K92Diseases of the digestive system7.4123.252210.22
L00–L98Diseases of the skin and subcutaneous tissue000.19131.0
M00–M99Diseases of the musculoskeletal and connective tissue001.10751.0
N00–N98Diseases of the genitourinary system3.7012.251530.46
R00–R99Symptoms, signs and abnormal clinical findings and lab. nos000.69471.0
V01–Y89External causes of morbidity and mortality7.4124.182840.31
Total  27 6802 
Table 2. Underlying cause of death (UCD) as reported by specific ICD-10 codes for subjects in the bullous pemphigoid (BP) group and the Centers for Disease Control (CDC) group
 UCD by specific ICD-10 codesBP group, % n CDC group, % n P-value
  1. ICD-10, International Classification of Diseases, 10th Revision.

F03Unspecified dementia18.5258.095500.063
I25.9Chronic ischemic heart disease, unspecified11.1139.046150.73
Total  27 6802 

Comparisons of MCD data revealed differences between the groups in several ICD-10 blocks. The control group had a higher rate of neoplasms (P = 0.025) and diseases of the circulatory system (P = 0.026). Conversely, higher rates of mental and behavioral disorders (P = 0.002), diseases of the digestive system (P = 0.025), and diseases of the genitourinary system (P = 0.025) were observed in the study group.

Specific MCD differences of statistical significance were found for sepsis, A41.9 (P = 0.031), dementia, F03 (P = 0.049), and major depressive disorder, F32.9 (P = 0.005).

Of the four patients with sepsis, UCDs were, respectively, cholangitis, pneumonia, urinary tract infection, and unknown. Only one of these patients was receiving immunosuppressant medication for BP at the time of death (Tables 3 and 4).

Table 3. Multiple causes of death (MCD) as reported by ICD-10 block for subjects in the bullous pemphigoid (BP) group and the Centers for Disease Control (CDC) groupa
 MCD by ICD-10 blockBP group, % n CDC group, % n P-value
  1. a

    No MCD data were recorded in the study or control group in the following ICD-10 blocks: O00–O91 Pregnancy, childbirth, puerperium; P00–P96 Certain conditions originating in the perinatal period, and U00–U99 Codes for special purposes.

  2. ICD-10, International Classification of Diseases, 10th Revision.

A00–B99Certain infections and parasitic disease3.7441.794520.13
C00–D48Neoplasms3.74410.3726230.025
D50–D89Diseases of the blood and blood-forming organs and certain disorders involving the immune system0.9311.142871.0
E00–E88Endocrine, nutritional and metabolic diseases8.4196.1415530.33
F01–F99Mental and behavioral disorders13.01145.9715100.002
G00–G98Diseases of the nervous system2.8034.0410220.80
H00–H57Diseases of the eye and adnexa000.29731.0
H60–H93Diseases of the ear and mastoid process000.04111.0
I00–I99Diseases of the circulatory system26.172836.5992510.026
J00–J98Diseases of the respiratory system10.281111.3028570.74
K00–K92Diseases of the digestive system7.4883.278270.025
L00–L98Diseases of the skin and subcutaneous tissue0.9310.38960.34
M00–M99Diseases of the musculoskeletal and connective tissue1.8722.536401.0
N00–N98Diseases of the genitourinary system9.35104.7311970.025
Q00–Q99Congenital malformations, deformations and chromosomal abnormalities000.13321.0
R00–R99Symptoms, signs and abnormal clinical findings and lab. nos1.8724.3110900.33
S00–T98Injury, poisoning and certain other consequences of other causes4.6753.508850.43
V01–Y89External causes of morbidity and mortality4.6753.488790.43
Total  107 25 285 
Table 4. Multiple causes of death (MCD) as reported by specific ICD-10 codes for subjects in the bullous pemphigoid (BP) group and the Centers for Disease Control (CDC) group
 MCD by specific ICD-10 codesBP group, % n CDC group, % n P-value
  1. ICD-10, International Classification of Diseases, 10th Revision.

A41.9Sepsis, unspecified organism3.7441.102770.031
F03Unspecified dementia8.4194.2610760.049
F32.9Major depressive disorder, single episode3.7440.631600.005
I10Essential (primary) hypertension4.6755.8014670.62
I25.1Atherosclerotic heart disease of native coronary artery3.7446.0515290.32
I50Heart failure3.7444.0510231.0
J18.9Pneumonia, unspecified organism4.6752.817100.23
Total  107 25 285 

A comparison of all infections listed within the MCD category is shown in Table 5. A specific difference of statistical significance was noted for infection as a contributing cause of death (P = 0.035).

Table 5. Multiple causes of death (MCD) as reported by specific ICD-10 codes pertaining to infectious etiologies for subjects in the bullous pemphigoid (BP) group and the Centers for Disease Control (CDC) group
 MCD by specific ICD-10 codes for infectionsBP group, % n CDC group, % n P-value
  1. ICD-10, International Classification of Diseases, 10th Revision.

J18.9Pneumonia, unspecified organism4.6752.817100.23
A41.9Septicemia, unspecified organism3.7441.102770.031
N39.0Urinary tract infection, site unspecified1.8720.721830.18
J18.0Bronchopneumonia, unspecified000.33831.0
I38Endocarditis, valve unspecified000.26661.0
A04.7Enterocolitis caused by Clostridium difficile000.09221.0
L03.9Cellulitis, unspecified000.06161.0
B18.2Chronic viral hepatitis C000.06141.0
A49.0Staphylococcal infection, unspecified000.05121.0
A49.9Bacterial infection, unspecified000.04101.0
B17.1Acute hepatitis C000.04101.0
I33.0Acute and subacute infective endocarditis000.04101.0
K83Cholangitis0.931 Not reported
Total 11.21125.5914130.012
Total minus K83 10.28115.5914130.035
Total  107 25 285 

Discussion

This study adds to current understanding of the factors contributing to death in patients with BP. Significant differences between the study and control groups were noted for several MCD ICD-10 blocks, including those pertaining to mental and behavioral disorders, diseases of the circulatory system, diseases of the digestive system, and diseases of the genitourinary system. Additionally, increased rates of sepsis, infection, dementia, and major depression as contributory causes of death were noted amongst the study group when specific ICD-10 codes were compared against those for the control group.

The association of BP and neurological disease has been studied extensively. A recent prospective study found degenerative neurological disease to be an independent risk factor for the development of BP.[15] Importantly, major depressive disorder emerged as a statistically significant cause of death upon review of the data. Although a previous diagnosis of dementia may be a confounder of these findings, the presence of BP among other immunobullous disease has been documented to decrease quality of life in patients with dementia.[16, 17] Therefore, quality of life and mental health should be an important consideration in the management of BP patients.

The association of sepsis and BP has been studied and remains an area of controversy.[5, 18] The present study systematically compared data for BP patients with CDC control data and found an increased rate of infection in the study group. Bullous pemphigoid, as well as other autoimmune bullous dermatoses, is associated with an increased risk for infection.[19] The elevated rates of infection and septicemia within the MCD category imply that infection control is an important consideration in the management of patients with autoimmune bullous disease.[19]

The association between BP and diseases of the circulatory system, especially stroke, has been discussed extensively.[4, 8-10] Such findings may be related to chronic inflammation and corticosteroid use as contributing factors in atherogenesis.[20] Increased disease of the circulatory system was observed in the control population. Our findings are not consistent with those of previous research. This may be the result of evaluating all diseases of the circulatory system, including diagnoses such as hypertension, stroke, and myocardial infarction, in the same ICD-10 block. When incidences of these individual diagnoses were compared between the study and control groups, no significant differences were noted between the groups for diseases of the circulatory system.

The study group demonstrated a lower rate of neoplasms and a higher rate of digestive and genitourinary system disease compared with the control population. Upon further examination of the specific ICD-10 codes, five of eight recorded MCD codes were suggestive of renal failure (N17.8 [n = 1], N18.9 [n = 2], N19 [n = 2]). No disease pattern was apparent amongst ICD-10 codes for digestive disease. Why these trends exist is unclear but would be worthy of future study.

This study has several limitations. Firstly, this was a retrospective study of a small number of patients, the majority of whom were women. As the USA does not maintain a national database on immunobullous disease, the number of patients available to our study was limited. Prior reports that have included large numbers of patients with BP have looked at rates of mortality but not at its causes. Our study is unique in that it evaluated a population of BP patients with death certificate data and therefore provides the most detailed mortality information reported to date. Additionally, the validity of death certificate data has been questioned previously[21] because cause(s) of death may be incorrectly reported on death certificates. However, a recent study demonstrated that death certificate data from the REP is sufficiently sensitive and specific to support an evaluation of certain mortality-based outcomes.[22]

The decision to study patients from Olmsted County, Minnesota, offered several unique advantages to this research project, including access to exceptional longitudinal subject follow-up, access to full chart data across specialties, excellent information on mortality, and complete follow-up capability for mortality with REP death data.[23] Additionally, age- and sex-specific mortality rates in Olmsted County have been found to be similar to state and national mortality rates.[24]

The relationship between BP and an increased mortality rate has been the subject of numerous studies; however, reports of cause of death data are scarce in the literature. We have demonstrated that there are significant associations between BP and dementia, infection, sepsis, and major depression as contributors to death. Future larger studies are needed to confirm these findings.

Acknowledgment

We are indebted to Barb Abbott, data retrieval specialist, Department of Epidemiology, Mayo Clinic, Rochester, MN, who assisted in the identification of study subjects.

Ancillary