Skip to main content
NCBI home page
JAMA Network logoLink to JAMA Network
. 2021 Mar 17;157(4):1–11. doi: 10.1001/jamadermatol.2020.5598

Mixed Individual-Aggregate Data on All-Cause Mortality in Bullous Pemphigoid

A Meta-analysis

Billal Tedbirt 1,2,✉, André Gillibert 3, Emilie Andrieu 1,2, Vivien Hébert 1,2, Sarah Bastos 1,2, Neil J Korman 4, Mark B Y Tang 5, Jun Li 6, Luca Borradori 7, Begonia Cortés 8, Soo-Chan Kim 9, Adrià Gual 10, Ting Xiao 11,12, Carilyn N Wieland 13, Janet A Fairley 14, Khaled Ezzedine 15, Pascal Joly 1,2
PMCID: PMC7970384  PMID: 33729430

Key Points

Question

Is bullous pemphigoid (BP) associated with increased mortality relative to the general population?

Findings

This meta-analysis combining individual and aggregate data from 56 unique studies of 12 340 unique patients found that patients with BP had a 2.93-fold increased 1-year mortality rate compared with the general population. Mortality was associated with BP itself, adverse effects of treatment, and/or patients’ comorbidities, with infections as the main cause of death.

Meaning

These findings may help clinicians in the management of BP and suggest avoiding aggressive treatments.

This meta-analysis estimates the 1-year standardized mortality ratio among patients with bullous pemphigoid.

Abstract

Importance

The 1-year standardized mortality ratio (SMR) of bullous pemphigoid (BP) has been reported as 2.15 to 7.56 and lower in the US than in Europe.

Objective

To estimate the worldwide 1-year SMR of BP.

Data Sources

PubMed, Embase, Cochrane Library, Google Scholar, Lissa, and gray literature (eg, medRxiv) were screened for studies of BP published from inception to June 10, 2020, with review of reference lists.

Study Selection

Retrospective and prospective studies reporting 1-year all-cause mortality rate in patients with BP and providing age statistics (eg, mean [SD]).

Data Extraction and Synthesis

Two reviewers independently extracted the data. The 1-year SMR was computed in studies reporting 1-year mortality by combining information on age obtained from studies with aggregate data and individual data. Risk of representativity, misclassification, and attrition bias were assessed by a custom tool.

Main Outcomes and Measures

The primary end point was the worldwide 1-year SMR. Secondary analysis included comparison of 1-year SMRs between continents in a meta-regression.

Results

Three studies were performed in the US (n = 260), 1 in South America (n = 45), 16 in Asia (n = 1903), and 36 in Europe (n = 10 132) for a total of 56 unique studies and 12 340 unique patients included in the meta-analysis (mean [SD] age, 77.3 [12.7] years; 55.9% women). The mean (SD) patient age in the United States was 75.6 (13.7) years; in Asia, 73.8 (13.6) years; and in Europe, 78.1 (12.3) years. The worldwide 1-year SMR was estimated at 2.93 (95% CI, 2.59-3.28; I2 = 85.6%) for all 56 studies. The 1-year SMR in the US was 2.40 (95% CI, 0.89-3.90; I2 = 86.3%) for 3 studies; in Asia, 3.53 (95% CI, 2.85-4.20; I2 = 86.3%) for 16 studies; and in Europe, 2.77 (95% CI, 2.35-3.19; I2 = 86.3%) for 36 studies. After adjustment on the expected 1-year mortality rate, the European 1-year SMR did not differ significantly from the 1-year SMR in the United States (−0.48 vs Europe; 95% CI, −2.09 to 1.14; P = .56) and Asia (0.51 vs Europe; 95% CI, −0.56 to 1.58; P = .35). Risk of attrition bias was high (>10% censorship) in 16 studies (28.6%), low in 16 (28.6%), and unclear in 24 (42.9%). Only 4 studies (7.1%) had a sampling method guaranteeing the representativity of BP cases in a population.

Conclusions and Relevance

Although heterogeneity was high and overall quality of follow-up was poor, this meta-analysis confirms the high mortality rate among patients with BP.

Introduction

Bullous pemphigoid (BP) is the most common autoimmune subepidermal blistering disease of the skin, which mainly affects elderly patients with frequent neurological comorbidities.1 Bullous pemphigoid is associated with significant morbidity and an increased death rate that seems to differ throughout the world. In Europe, the 1-year death rate ranged from 0 to 52%,2,3 whereas it ranged from 11% to 23% in the US4,5 and from 0 to 42% in Asia.6,7

Some authors prefer to compute a standardized mortality ratio (SMR) to assess the extent to which the mortality of patients with BP is higher than the mortality due to diseases that come with old age in this population. The all-cause SMR of BP is the ratio between the observed all-cause mortality rate in patients with BP and the expected all-cause mortality rate in the general population with the same age and sex distribution as patients with BP. The SMR can be interpreted as a relative risk of death during a given period (eg, 1 year) compared with the general population. An SMR of greater than 1.00 means that there is an increased mortality rate relative to the general population, due to the disease (ie, BP), its treatments, or comorbidities. Major differences in the all-cause SMR of patients with BP have been reported from study to study and country to country, ranging from 2.15 to 7.568,9 for 1-year SMR and from less than 1.00 to 6.60 for global SMR.4,10

Older age, low Karnofsky score, hypoalbuminemia, high dose of oral corticosteroids, and neurological comorbidities (dementia, Parkinson disease, and stroke) have been reported to be associated with mortality in European series,11,12,13 whereas the absence of hospitalization of patients with BP in the US has been suggested to explain the lower death rate of patients with BP in 1 US study.5 To disentangle the discrepancies in the all-cause mortality of patients with BP among the different studies throughout the world, we performed a meta-analysis combining individual patient and aggregate data to estimate the all-cause 1-year SMR of patients with newly diagnosed BP (primary analysis), assess its heterogeneity, and compare it between continents.

Methods

This systematic review and meta-analysis was registered on PROSPERO.14 We followed the Meta-analysis of Observational Studies in Epidemiology (MOOSE)15 and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines.16 Ethics committee approval was not required for this meta-analysis of published data.

Eligibility Criteria

Inclusion criteria were either prospective or retrospective studies including patients with BP, for which the 1-year overall death rate could be extracted, and age distribution information of patients with BP was provided with at least 1 position statistic (eg, mean or median) or frequency by age class (eg, 20-40 years, 40-60 years, etc). Exclusion criteria were (1) massive inclusion (≥50%) of prevalent cases; (2) inclusion of specific BP populations that may have a different prognosis (eg, cancer, refractory BP); (3) inclusion of patients with other pemphigoid without possibility of isolating patients with BP; (4) reports of only cause-specific mortality; (5) case reports; and (6) systematic reviews. There was no exclusion based on methodology quality, diagnosis method, or number of patients.

Information Sources and Literature Search

We performed a comprehensive systematic search in 15 different languages using PubMed, Embase, Cochrane Library (Cochrane Central Register of Controlled Trials and Cochrane Database of Systematic Reviews), Google Scholar, Lissa, and gray literature (http://www.theses.fr and medRxiv) to screen studies published from inception to June 10, 2020 (first search in January 2020; updated in June 2020). The main search terms were pemphigoid AND mortality, death, lethality, survival, or prognosis, as described in eTable 1 in the Supplement. All prospective and retrospective studies were included with no language restriction, and the reference lists in relevant review articles were scanned manually as well. Two authors of this study (B.T. and A.G.) screened all eligible articles based on the title, abstract, and full text, with disagreements identified by software and solved by consensus (data screening in the eMethods in the Supplement).

Data Extraction

Data extraction (eMethods in the Supplement) was conducted independently by 2 authors (B.T. and A.G.), with disagreements solved by consensus. The general characteristics of the studies were recorded: inclusion period, study area, and whether the follow-up of vital status was obtained passively (from medical records), actively (telephone calls or email to patients or family physician), or exhaustively (consultation of death registry). In addition, the mode of patient’s inclusion (incident only or prevalent and incident), total number of patients, and sex ratio when available were recorded. The following statistics about age distribution were retrieved if available: mean, SD, minimum, maximum, median, quartiles, and frequency by age class (eg, 18-60 years, 60-79 years, etc). Sex-specific statistics were retrieved if available. The distribution of inclusion dates (eg, by 2-year periods) was recorded when available.

If possible, the Kaplan-Meier method was used for the estimation of the 1-year death rate (with graphical extraction from figures and pooling subgroups, if needed). Otherwise, the number of deaths, survivors, and censorships at 1 year were recorded and the death rate was computed with a modified Kaplan-Meier estimator, assuming a uniform distribution of censorships (computation of SMRs in the eMethods in the Supplement).

Frequencies of each cause of death were extracted, counting either causes of deaths of all patients dying during the first year after BP diagnosis or causes of deaths of patients dying at any follow-up time that could be longer or shorter than 1 year or both. Causes of death were not extracted if they were only reported in a subgroup, such as treatment-related deaths.

Individual patient data were searched for 12 studies4,5,10,17,18,19,20,21,22,23,24,25 in a previous meta-analysis.26 The authors, contacted by email, returned deidentified individual data regarding birthdate, sex, date of diagnosis, date of death, and optionally cause of death. In addition, published data of 4 studies containing detailed individual data tables were considered as individual data studies.2,27,28,29 Risk of bias was assessed using a 3-dimensional custom tool created for this meta-analysis, remotely based on Newcastle-Ottawa Scale,30 but adapted to noncomparative studies, to assess representativity, misclassification, and attrition bias (eMethods in the Supplement).

Statistical Analysis

Computation of SMRs for Each Study

The SMRs have been computed as the rate between the crude 1-year death rate and the expected death rate for a general population of the same country, period, and age pyramid. For individual data studies, the exact age pyramid was available. For aggregate data studies, an approximate age pyramid was estimated by distortion of all individual-data studies pooled together until it fit the aggregate statistics. A nonlinear spline-based model fitted on the individual patient data was used to describe the association of the sex-ratio with age (eMethods in the Supplement). In aggregate data studies, the age-sex distribution was derived from this nonlinear model and from the global sex-ratio of the study. The United Nations mortality table of each nation was retrieved in October 2019 from the estimates and standard projection variants data sets.31 The 5-year precise age-, sex-, and period-specific death rates were interpolated to get 1-year precision assuming a local Gompertz distribution (eMethods in the Supplement).32 With the 1-year death rate, joint age-sex distribution, distribution of inclusions and country of inclusions, the SMR was computed (eMethods in the Supplement).

Estimation and Comparison of Pooled SMRs

The worldwide SMR (primary analysis) was pooled in a random-effects linear model without transformation. Continent-specific death rates and SMRs (secondary preplanned analyses) were estimated in single mixed-effects meta-regression DerSimonian-Laird linear models. A comparison of SMRs between continents was performed in a meta-regression with linear adjustment on the expected 1-year death rate of the study (denominator of the SMR, quantitative variable). For subgroup analyses (eg, by age class), the variance-stabilizing square root transformation was applied to the random-effects model, to adapt to small samples with a strong covariance between variance and the proportion (post hoc decision).33

Estimation and Comparisons of Crude Death Rates and Directly Standardized Rates

The worldwide crude 1-year death rate was pooled (post hoc analysis) in a random-effects linear model, without transformation. The eMethods in the Supplement provides details of the post hoc multivariate analyses of factors associated with death rates. Death rates directly standardized on the pooled age-sex distribution of individual data studies were computed in each individual data study (eMethods in the Supplement).

Long-term Mortality and Causes of Deaths

Long-term mortality was assessed by the crude (unadjusted) annualized death rates by period (1-2 years, 2-3 years, and 3-5 years) after the inclusion (post hoc analysis) (eMethods in the Supplement). Causes of death were grouped (eg, pneumopathy into infectious diseases) and described by proportions (eMethods in the Supplement).

Sensitivity Analyses and Analysis of Publication Bias

Because we noticed that 2 studies with passive follow-up had very low SMRs,3,6 significantly below 1.00, we performed a post hoc sensitivity analysis excluding these 2 studies with the same model as the primary analysis. We assessed the publication bias by a funnel plot by a graphical interpretation. No multiple testing procedures were used. All analyses were performed in R statistical software, version 3.6 (R Foundation for Statistical Computing). All tests were 2-sided Wald tests at the 5% significance level.

Results

Study Selection and Characteristics

The search strategy led to a total of 1488 unique citations eligible after title and abstract screening. After screening, 77 full-text articles were read, and 56 were included in the analysis (Figure 1). Among the 56 unique articles2,3,4,5,6,7,8,9,10,11,12,17,18,19,20,21,22,23,24,25,27,28,29,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66 (12 340 unique patients), individual patient data were available for 16 studies (n = 2122).2,4,5,10,17,18,19,20,21,22,23,24,25,27,28,29 Characteristics of included studies, published from 1978 to April 2020, are summarized in Table 1.

Figure 1. Flowchart According to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) for the Systematic Literature Search.

Figure 1.

Open in a new tab

Table 1. Characteristics of the Included Studies.

Source Country Period Diagnostic method Censorship at 1 y, % Incident only Follow-up No. of patients (No. female) Estimated age, mean (SD), y Age distribution data
Bastos et al,17 2019 7a 2015-2018 Unknown 8.0 Yes Passive 187 (112) 81.5 (8.6) Individual data
Monshi et al,34 2020 Austria 2001-2012 IF 8.0 Yes Passive 100 (56) 79.7 (9.7) Position, dispersion
Sticherling et al,35 2017 Austria, Germany 2001-2005 IF 18.5 No Active 54 (39) 77.0 (13.7) Position, dispersion
Kibsgaard et al,36 2017 Denmark 1977-2015 COD 0.0 Yes Exhaustive 3281 (1834) 76.5 (12.6) Histogram
Försti et al,9 2016 Finland 1985-2012 IF 0.0 Yes Exhaustive 198 (102) 77.5 (10.4) Position, dispersion
Bernard et al,37 1986 France 1973-1982 IF Unknown Yes Passive 57 (31) 75.4 (10.7) Position, range
De Rocco,2 1986 France 1975-1984 IF 5.8 Yes Active 103 (57) 75.5 (9.2) Individual data
Devendeville et al,28 1990 France 1978-1988 IF 26.7 No Active 30 (14) 81.0 (6.5) Individual data
Michel et al,38 1999 France 1981-1995 IF >19.4 Yes Passive 62 (33) 77.0 (10.0) Position, range
Taïeb et al,27 1986 France 1982-1984 IF 0.0 Yes Active 10 (4) 74.1 (7.8) Individual data
Roujeau et al,39 1998 France 1985-1992 IF 3.7 Yes Active 217 (120) 79.0 (11.0) Position, dispersion
Depaire,29 1995 France 1993-1995 IF 20.0 No Active 20 (14) 80.5 (13.7) Individual data
Joly et al,18 2002 France 1996-1998 IF 38.1 Yes Active 341 (214) 81.4 (9.6) Individual data
Chevalier et al,40 2016 France 1997-2011 IF Unknown Yes Passive 178 (122) 79.5 (11.8) Position, dispersion
Joly et al,10 2012 France 2001-2004 IF 35.9 Yes Passive 198 (121) 82.3 (9.7) Individual data
Nespoulous et al,41 2018 France 2004-2017 Unknown Unknown Yes Unknown 329 (187) 83.3 (8.7) Position
Cordel et al,42 2009 France 2006-2009 IF Unknown Yes Active 26 (17) 86.3 (8.5) Position, dispersion
Cantegrit et al,43 2019 France 2010-2015 IF Unknown Yes Passive 67 (46) 82.0 (8.4) Position, dispersion
Clapé et al,44 2020 France 2013-2017 Unknown 8.4 Yes Active 95 (60) 81.8 (9.4) Position, dispersion
Rzany et al,11 2002 Germany 1987-1997 IF <8.1 Yes Active 369 (199) 77.8 (11.1) Position, dispersion
Kyriakis et al,3 1999 Greece 1987-1988 IF Unknown No Unknown 27 (13) 74.8 (10.6) Position, dispersion
Serwin et al,45 2014 Poland 1999-2012 IF <41.0 Yes Passive 122 (70) 74.2 (12.1) Position, dispersion
Kalinska-Bienias et al,46 2017 Poland 2000-2013 IF 0.0 Yes Exhaustive 205 (131) 76.2 (11.8) Histogram
Gual et al,19 2014 Spain 1989-2010 IF 10.9 Yes Passive 101 (49) 77.7 (11.5) Individual data
Garcia-Doval et al,8 2005 Spain 1998-2003 IF 15.4 Yes Active 26 76.5 (11.3) Position, range
Heilborn et al,47 1999 Sweden 1996-1997 IF 0.0 Yes Active 11 (7) 80.3 (8.2) Position, range
Kjellman et al,48 2008 Sweden 1999-2003 IF Unknown Yes Passive 138 (79) 81.0 (10.6) Position, range
Fisch et al,49 2018 Sweden 2006-2015 Unknown Unknown Yes Unknown 232 (122) 84.6 (6.7) Position, dispersion
Cortés et al,50 2012 Switzerland 1990-2003 IF Unknown Yes Active 60 (34) 79.5 (11.6) Position, dispersion
Cortés et al,20 2011 Switzerland 2001-2002 IF 20.0 Yes Passive 113 (66) 78.1 (12.4) Individual data
Burton et al,51 1978 UK 1973-1973 IF Unknown Yes Active 25 (16) 74.8 (14.0) Position
Venning and Wojnarowska,12 1992 UK 1975-1988 IF 15.9 Yes Passive 82 (44) 73.9 (11.0) Position, range
Gudi et al,52 2005 UK 1991-2001 IF Unknown Yes Passive 83 (50) 79.2 (12.8) Histogram
Persson et al,53 2020 UK 1998-2017 COD 0.0 Yes Exhaustive 2658 (1497) 77.9 (13.9) Histogram
Angit et al,54 2011 UK 2006-2007 IF 23.0 Yes Passive 74 (39) 79.7 (9.4) Position, range
Chalmers et al,55 2017 UK and Germany 2009-2013 IF 23.3 No Active 253 (120) 77.7 (9.7) Position, dispersion
Brick et al,21 2014 US 1963-2009 IF 4.9 Yes Passive 82 (46) 77.7 (12.0) Individual data
Colbert et al,5 2004 US 1997-2002 IF 18.4 Yes Passive 32 (15) 77.0 (12.2) Individual data
Parker et al,4 2008 US 1998-2003 IF 0.0 Yes Exhaustive 146 (89) 74.2 (14.6) Individual data
Li et al,22 2013 China 1991-2011 IF Unknown Yes Active 140 (58) 64.7 (13.5) Individual data
Zhang et al,23 2013 China 2005-2010 IF 23.4 Yes Passive 94 (41) 71.1 (12.7) Individual data
Chang,7 2013 Hong Kong 2002-2011 IF Unknown Yes Unknown 121 79.9 (9.9) Position
Mokhtari et al,56 2019 Iran 2008-2016 IF 36.2 Yes Passive 69 (41) 69.6 (13.6) Position, dispersion
Kridin et al,57 2019 Israel 2000-2015 IF 0.0 Yes Exhaustive 287 (168) 77.6 (12.1) Position, dispersion
Rozenblat et al,58 2019 Israel 2009-2016 IF Unknown Yes Active 87 (40) 79.6 (9.1) Histogram
Kanamori et al,59 2004 Japan 1988-2003 Unknown Unknown Yes Unknown 26 (15) 73.1 (14.3) Position, range
Sasai et al,60 2015 Japan 2003-2012 Unknown Unknown Yes Passive 52 (29) 84.3 (6.3) Position, range
Lee and Kim,24 2014 South Korea 1993-2013 IF Unknown Yes Passive 166 (83) 69.9 (14.3) Individual data
Jeon et al,61 2018 South Korea 2006-2013 IF Unknown Yes Active 103 (50) 74.4 (10.6) Histogram
Nanda et al,62 2006 Kuwait 1991-2005 IF 14.0 Yes Passive 43 (36) 65.2 (18.8) Histogram
Phoon et al,63 2015 Singapore 2002-2011 IF Unknown Yes Passive 97 (48) 79.0 (11.0) Position, dispersion
Cai et al,25 2014 Singapore 2004-2009 IF 0.0 Yes Exhaustive 359 (187) 75.7 (12.6) Individual data
Wei,64 2018 Taiwan 2012-2017 IF Unknown Yes Unknown 163 (78) 70.0 (12.3) Position, dispersion
Kulthanan et al,65 2011 Thailand 1991-2009 IF Unknown Yes Passive 58 (42) 69.3 (14.7) Position, dispersion
Kızılyel et al,6 2015 Turkey 2003-2013 IF Unknown Yes Passive 38 (23) 62.4 (21.0) Position, dispersion
Carvajal Aguilera et al,66 2020 Chile 2005-2017 IF Unknown Yes Passive 45 (24) 72.2 (16.6) Position, dispersion
Open in a new tab

Abbreviations: COD, code from a medical dictionary (eg, International Statistical Classification of Diseases and Related Health Problems, Tenth Revision); IF, direct or indirect immunofluorescence plus standard histologic evaluation.

a

Includes France, the Netherlands, Greece, Bulgaria, Poland, Germany, and Czech Republic.

The meta-analysis included 36 studies (Figure 2) performed in Europe (n = 10 132), including 14 in France (n = 1733),2,10,18,27,28,29,37,38,39,40,41,42,43,44 5 in the UK (n = 2922),12,51,52,53,54 3 in Sweden (n = 381),47,48,49 and 1 in Denmark (n = 3281).36 A total of 16 studies (n = 1903) were performed in Asia,6,7,22,23,24,25,56,57,58,59,60,61,62,63,64,65 including Turkey, Iran, Kuwait, Israel, Taiwan, China, Hong Kong, Singapore, Japan, Thailand, and South Korea. Four studies were included from North and South America: 3 performed in the US (n = 260)4,5,21 and 1 in Chile (n = 45).66 The estimated mean (SD) ages, after pooling studies, were 75.6 (13.7) years in the US, 73.8 (13.6) years in Asia, and 78.1 (12.3) years in Europe. Women represented 55.9% and men represented 44.1% of the 12 340 patients (mean [SD] age, 77.3 [12.7] years).

Figure 2. Geographic Distribution of Patients With Bullous Pemphigoid at Inclusion in Studies Included in the Meta-analysis.

Figure 2.

Open in a new tab

Estimation and Comparison of Pooled SMRs

The worldwide 1-year SMR (Figure 3) was estimated at 2.93 (95% CI, 2.59-3.28; I2 = 85.6%; 56 studies). The τ value was estimated at 1.13, interpreted as the between-study SD of the true SMR. Therefore, the typical difference of SMR between 2 studies was estimated at 1.13, such as would be found for one study with an SMR at 2.00 and another with an SMR at 3.13. The coefficient of variation of the true SMR from study to study was estimated at 1.13/2.93 = 38.5%. The SMR was estimated at 2.77 (95% CI, 2.35-3.19; 36 studies) in Europe, 2.40 (95% CI, 0.89-3.90; 3 studies) in the US, 3.53 (95% CI, 2.85-4.20; 16 studies) in Asia, and 2.03 (95% CI, 0.33-3.74; 1 study) in Chile, with a pooled residual heterogeneity I2 = 86.2%. After adjustment with the expected 1-year death rate in the study (multivariate meta-regression), the difference between SMR for the US compared with Europe was estimated at −0.48 (95% CI, −2.09 to 1.14; P = .56); between Asia and Europe, 0.51 (95% CI, −0.56 to 1.58; P = .35).

Figure 3. Forest Plot of All-Cause 1-Year Standardized Mortality Ratio (SMR).

Figure 3.

Open in a new tab

Different sizes of markers indicate different weights. Wide points of diamonds indicate SMR; ends of diamonds, 95% CI.

aIncludes France, the Netherlands, Greece, Bulgaria, Poland, Germany, and Czech Republic.

Estimation and Comparison of Crude Death Rates

The crude 1-year death rate from the 56 studies (eFigure 1 in the Supplement) ranged from 0 to 52.4%.2,3,6 The worldwide crude 1-year death rate (primary analysis) was estimated at 22.0% (95% CI, 19.4%-24.6%; I2 = 89.5%; 56 studies). In Europe, the crude 1-year death rate was estimated at 23.1% (95% CI, 19.9%-26.4%; 36 studies); in the US, 17.7% (95% CI, 6.5%-28.9%; 3 studies); in Asia, 20.8% (95% CI, 16.0%-25.6%; 16 studies); and in Chile, 11.1% (95% CI, 1.8%-20.4%; 1 study). Pooled residual heterogeneity was I2 = 89.2%. Other post hoc meta-regression models are described in the eResults in the Supplement.

Individual Data Analysis

Sixteen studies2,4,5,10,17,18,19,20,21,22,23,24,25,27,28,29 had individual data available that were analyzed. Of these studies, 9 were conducted in Europe (n = 1103),2,10,17,18,19,20,27,28,29 3 in the US (n = 260),4,5,21 and 4 in Asia (n = 759).22,23,24,25 Of the 9 studies in Europe, 6 (n = 702) were conducted in France.2,10,18,27,28,29 The mean (SD) age of all individual data studies pooled together was 76.7 (12.7) (median, 78.7 [range, 18.9-104.3]) years, with a male-to-female sex ratio of 0.814. The precise age and sex distribution are shown in eFigure 2 in the Supplement. The worldwide 1-year random-effects SMR estimated from these 16 studies was 3.02 (95% CI, 2.37-3.75, I2 = 86.0%). The SMRs in age subgroups are described in Table 2. The random-effects SMR was not significantly different (P = .78) between men (SMR, 3.00; 95% CI, 2.35-3.73; I2 = 70.2%; 16 studies) and women (SMR, 3.17; 95% CI, 2.34-4.12; I2 = 82.7%; 16 studies). Directly standardized death rates are described in eFigure 3 and the eResults in the Supplement.

Table 2. All-Cause 1-Year SMR and Absolute Risk Differences of Bullous Pemphigoid by Agea.

Variable Age group, y
<60 (n = 210) 60-69 (n = 297) 70-79 (n = 631) 80-89 (n = 758) >90 (n = 226)b
1-y death rate, No. (%)
Expected 0.9 (0.42) 5.1 (1.7) 24.8 (3.93) 72.5 (9.55) 45.7 (20.22)
Observed 15 (7.14) 40 (13.47) 121 (19.18) 224 (29.55) 88 (38.94)
Absolute risk difference estimate (95% CI), %
Fixed effects 6.7 (3.2-10.2) 11.8 (7.9-15.7) 15.2 (12.2-18.3) 20.0 (16.7-23.2) 18.7 (12.4-25.0)
Random effects 10.2 (6.1-15.2) 13.5 (9.7-17.8) 14.1 (9.0-20.1) 21.8 (13.4-31.4) 18.8 (8.6-30.6)
SMR (95% CI), %
Fixed effects 17.4 (9.1-25.8) 8.0 (5.7-10.3) 4.9 (4.1-5.7) 3.1 (2.8-3.4) 1.9 (1.6-2.2)
Random effects 25.4 (15.2-38.1) 8.9 (6.1-12.1) 4.6 (3.2-6.1) 3.1 (2.4-4.0) 1.9 (1.4-2.4)
Open in a new tab

Abbreviation: SMR standardized mortality ratio.

a

Includes the 16 studies where individual data are available. Data are estimated in random effects (planned analysis) and fixed effects (sensitivity analysis) models.

b

Includes 6 patients aged 100 years or older.

Long-term Mortality

All-cause mortality beyond 1 year, assessed in large studies25,36,46,53,57 (≥200 patients) with follow-up of at least 3 years and exhaustive mortality reporting, is reported in eTable 2 in the Supplement. The death rates were always higher in the first year (20% to 27%) than in subsequent years (9% to 16%).

Causes of Death

A total of 690 deaths had their causes reported (eResults in the Supplement). The main causes of death (eTable 3 in the Supplement) were infectious diseases (44.4% at 1 year and 46.1% at any follow-up time) and cardiovascular diseases (33.2% at 1 year and 33.9% at any follow-up time).

Sensitivity Analyses and Risk of Bias

After exclusion of the 2 studies3,6 that had an SMR significantly less than 1.00, suggesting a high risk of underestimation of the true death rate, the worldwide 1-year SMR was estimated at 3.04 (95% CI, 2.75-3.34; I2 = 78.1%; 54 studies). There was no evidence of publication bias according to the funnel plot (eFigure 4 in the Supplement). Risk of attrition bias was high (>10% censorship) in 16 studies (28.6%), low in 16 (28.6%) and unclear in 24 (42.9%). The risk of representativity was low in 4 studies (7.1%); risk of misclassification, 48 studies (85.7%); and risk of attrition bias, 16 studies (28.6%) (eFigure 5 and eResults in the Supplement).

Discussion

This systematic review and meta-analysis is based on aggregate and individual patient data from 56 studies of patients with BP involving 12 340 patients from throughout the world, including Europe, Asia, the US, and Chile. We found a 2.93-fold higher risk of all-cause death at 1 year in patients with BP compared with the general population in the same countries after adjusting for age and sex in a meta-analysis with a high heterogeneity (I2 = 85.6%, τ = 1.13). There was no significant difference between continents.

Kridin et al57 performed a meta-analysis of death rates, finding a worldwide 1-year death rate similar to ours (23.5% vs 22.0%). With our combined aggregate-individual data, we could assess SMRs in all countries and compare SMRs between continents. Moreover, we analyzed all-cause mortality beyond 1 year on the most robust studies.

Roujeau et al39 found a high death rate (41% at 1 year). We confirm the fact that this study has one of the highest death rates (eFigure 1 in the Supplement) we have seen. Korman67 hypothesized that the study of Roujeau et al39 may have a selection bias, favoring the inclusion of severe generalized forms. Korman also hypothesized that ethnic differences could be the cause of higher death rates in France; we did not confirm this latter hypothesis, because most other French studies found much lower death rates.18,28,29,38,41,42,43,44

We showed a higher all-cause mortality during the first year of follow-up than thereafter. This finding suggests that the first-year mortality may be due to BP itself and adverse effects of treatment, whereas some part of the excess mortality beyond the first year of follow-up may be additionally due to associated chronic disorders, which are particularly frequent in patients with BP.7,13,40,46 As expected, infections and cardiovascular disorders, which are favored by the older age of patients with BP and the use of corticosteroids and immunosuppressive drugs, were the main cause of death observed in this meta-analysis.

As shown in Table 2, the 1-year SMR decreases greatly with age, although the absolute risk difference at 1 year increases with age. The very high SMR (>15.00) in younger patients (<60 years) with fewer associated disorders confirms the fact that BP or its treatment can be deadly. Older patients may either die of their BP (as demonstrated by the high absolute risk difference) or of associated disorders, as demonstrated by the high expected risk of death without BP and lower SMR. Because SMRs depend on age, they cannot be used to reliably compare the prognosis of BP in different populations with very different average ages. Instead, direct standardization should be performed.

Strengths and Limitations

An important strength of our study was the combination of individual and aggregate data, allowing inclusion of all studies reporting basic age statistics and 1-year crude death rate (56 studies), including the ones obtained by graphical extraction. However, our data were limited because there were few US studies, only 1 South American study, and no African studies. Many studies had a poor quality of follow-up, with 1-year loss to follow-up as high as 38%, and 2 studies had SMRs significantly less than 1.00, probably because patients who died were lost to follow-up.3,6 Beyond 1 year, mortality data were too scarce to be reliably included in the meta-analysis. The poor quality of follow-up may explain another part of the high heterogeneity we found (I2 = 85.6%). Because many patients were recruited at the hospital, with variable indications of hospitalizations, a differential selection bias may explain part of the heterogeneity. Although most studies recruited older patients, some of them recruited younger patients6,22,62 who have a different prognosis (lower death rate but higher SMR). Inclusion of inpatients could lead to a selection bias with an overestimation of the death rate, such as suggested by Chang.7 There was a high risk of misclassification bias in the 2 studies using registry linking by code from medical dictionaries.36,53 Moreover, the exact clinical, histological, and immunological diagnostic criteria varied from study to study and were often poorly reported, with a possible evolution with time (from 1963 to 2018).

Causes of death were recorded and reported from medical records without standardized methods in all articles except that of Cai et al,25 who used a national death registry. Even in the study by Cai et al,25 the death registry may include possible conventions and practices that differ from those of other countries. Moreover, causes of deaths were sometimes grouped without standardization. One article may include pneumopathy in respiratory causes, whereas another would include it in infectious diseases. When multiple causes of deaths were identified, only one of them was reported by authors; the selection method was never reported. Many causes of deaths (157 of 690 [22.8%]) were unknown in articles reporting causes of death, with probable selection bias (eg, fewer missing data in hospital deaths).

Conclusions

This meta-analysis found that patients with BP had a major increase in the risk of all-cause death in all continents, with a nonsignificant difference between continents. However, the low number of US studies cannot allow us to conclude that there is no difference.

Supplement.

eTable 1. Search Strategy in Online Databases

eTable 2. Long-term All-Cause Mortality Assessment

eTable 3. Causes of Death Extracted at 1-Year or Longer Follow-up

eMethods. Data Screening, Extraction, and Analysis

eFigure 1. Forest Plot of All-Cause 1-Year Mortality Rate

eFigure 2. Age Distribution at Inclusion in Pooled Individual Data Studies and Age-Sex Relationship

eFigure 3. Forest Plot of All-Cause Directly Standardized 1-Year Mortality Rates

eFigure 4. Funnel Plot of All-Cause 1-Year Death Rates of the 56 Studies Included in the Meta-analysis

eFigure 5. Assessment of Risk of Bias in the 56 Studies on a Custom Scale

eResults. Comparison of Crude Death Rates, Directly Standardized Death Rates, Causes of Death, and Risk of Bias

Click here for additional data file. (1.2MB, pdf)

References

  • 1.Bastuji-Garin S, Joly P, Lemordant P, et al. ; French Study Group for Bullous Diseases . Risk factors for bullous pemphigoid in the elderly: a prospective case-control study. J Invest Dermatol. 2011;131(3):637-643. doi: 10.1038/jid.2010.301 [DOI] [PubMed] [Google Scholar]
  • 2.De Rocco F. Pemphigoide bulleuse: devenir, facteurs pronostiques, traitement: a propos de 130 observations. Master’s thesis. Université Lille-II; 1986. [Google Scholar]
  • 3.Kyriakis KP, Paparizos VA, Panteleos DN, Tosca AD. Re-evaluation of the natural course of bullous pemphigoid: a prospective study. Int J Dermatol. 1999;38(12):909-913. doi: 10.1046/j.1365-4362.1999.00754.x [DOI] [PubMed] [Google Scholar]
  • 4.Parker SRS, Dyson S, Brisman S, et al. Mortality of bullous pemphigoid: an evaluation of 223 patients and comparison with the mortality in the general population in the United States. J Am Acad Dermatol. 2008;59(4):582-588. doi: 10.1016/j.jaad.2008.07.022 [DOI] [PubMed] [Google Scholar]
  • 5.Colbert RL, Allen DM, Eastwood D, Fairley JA. Mortality rate of bullous pemphigoid in a US medical center. J Invest Dermatol. 2004;122(5):1091-1095. doi: 10.1111/j.0022-202X.2004.22504.x [DOI] [PubMed] [Google Scholar]
  • 6.Kızılyel O, Elmas ÖF, Bilen H, et al. Bullous pemphigoid in Erzurum: a 10 year retrospective study. Turkderm. 2015; 49(1):66-69. doi: 10.4274/turkderm.12144 [DOI] [Google Scholar]
  • 7.Chang M. Bullous pemphigoid: a review of patients managed in hospital and its risk factors for poor prognosis and early mortality. Abstract presented at: Hong Kong Society of Dermatology and Venereology Annual Scientific Meeting; May 26, 2013; Hong Kong. [Google Scholar]
  • 8.Garcia-Doval I, Conde Taboada A, Cruces Prado MJ. Sepsis associated with dermatologic hospitalization is not the cause of high mortality of bullous pemphigoid in Europe. J Invest Dermatol. 2005;124(3):666-667. doi: 10.1111/j.0022-202X.2005.23628.x [DOI] [PubMed] [Google Scholar]
  • 9.Försti A-K, Jokelainen J, Timonen M, Tasanen K. Risk of death in bullous pemphigoid: a retrospective database study in Finland. Acta Derm Venereol. 2016;96(6):758-761. [DOI] [PubMed] [Google Scholar]
  • 10.Joly P, Baricault S, Sparsa A, et al. Incidence and mortality of bullous pemphigoid in France. J Invest Dermatol. 2012;132(8):1998-2004. doi: 10.1038/jid.2012.35 [DOI] [PubMed] [Google Scholar]
  • 11.Rzany B, Partscht K, Jung M, et al. Risk factors for lethal outcome in patients with bullous pemphigoid: low serum albumin level, high dosage of glucocorticosteroids, and old age. Arch Dermatol. 2002;138(7):903-908. doi: 10.1001/archderm.138.7.903 [DOI] [PubMed] [Google Scholar]
  • 12.Venning VA, Wojnarowska F. Lack of predictive factors for the clinical course of bullous pemphigoid. J Am Acad Dermatol. 1992;26(4):585-589. doi: 10.1016/0190-9622(92)70085-T [DOI] [PubMed] [Google Scholar]
  • 13.Joly P, Benichou J, Lok C, et al. Prediction of survival for patients with bullous pemphigoid: a prospective study. Arch Dermatol. 2005;141(6):691-698. doi: 10.1001/archderm.141.6.691 [DOI] [PubMed] [Google Scholar]
  • 14.National Institute for Health Research . Mortality in bullous pemphigoid: a systematic review and meta-analysis combining individual patient data and aggregate data. CRD42020164085. Accessed November 29, 2020. https://www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=164085
  • 15.Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting: Meta-analysis of Observational Studies in Epidemiology (MOOSE) group. JAMA. 2000;283(15):2008-2012. doi: 10.1001/jama.283.15.2008 [DOI] [PubMed] [Google Scholar]
  • 16.Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group . Preferred Reporting Items for Systematic Reviews and Meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535. doi: 10.1136/bmj.b2535 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Bastos S, Hebert V, Vassileva S, et al. Efficacy and tolerance of prednisone at 0.5mg/kg/day in initial treatment of bullous pemphigoid [in French]. Ann Dermatol Vénéréol. 2019;146(12, suppl):A61-A62. doi: 10.1016/j.annder.2019.09.038 [DOI] [Google Scholar]
  • 18.Joly P, Roujeau J-C, Benichou J, et al. ; Bullous Diseases French Study Group . A comparison of oral and topical corticosteroids in patients with bullous pemphigoid. N Engl J Med. 2002;346(5):321-327. doi: 10.1056/NEJMoa011592 [DOI] [PubMed] [Google Scholar]
  • 19.Gual A, Mascaró JM Jr, Rojas-Farreras S, Guilabert A, Julià M, Iranzo P. Mortality of bullous pemphigoid in the first year after diagnosis: a retrospective study in a Spanish medical centre. J Eur Acad Dermatol Venereol. 2014;28(4):500-506. doi: 10.1111/jdv.12065 [DOI] [PubMed] [Google Scholar]
  • 20.Cortés B, Marazza G, Naldi L, Combescure C, Borradori L; Autoimmune Bullous Disease Swiss Study Group . Mortality of bullous pemphigoid in Switzerland: a prospective study. Br J Dermatol. 2011;165(2):368-374. doi: 10.1111/j.1365-2133.2011.10413.x [DOI] [PubMed] [Google Scholar]
  • 21.Brick KE, Weaver CH, Lohse CM, et al. Incidence of bullous pemphigoid and mortality of patients with bullous pemphigoid in Olmsted County, Minnesota, 1960 through 2009. J Am Acad Dermatol. 2014;71(1):92-99. doi: 10.1016/j.jaad.2014.02.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Li J, Zuo Y-G, Zheng H-Y. Mortality of bullous pemphigoid in China. JAMA Dermatol. 2013;149(1):106-108. doi: 10.1001/archdermatol.2012.2994 [DOI] [PubMed] [Google Scholar]
  • 23.Zhang L-M, Wu J, Xiao T, et al. Treatment and mortality rate of bullous pemphigoid in China: a hospital-based study. Eur J Dermatol. 2013;23(1):94-98. doi: 10.1684/ejd.2012.1906 [DOI] [PubMed] [Google Scholar]
  • 24.Lee JH, Kim S-C. Mortality of patients with bullous pemphigoid in Korea. J Am Acad Dermatol. 2014;71(4):676-683. doi: 10.1016/j.jaad.2014.05.006 [DOI] [PubMed] [Google Scholar]
  • 25.Cai SCS, Allen JC, Lim YL, Chua SH, Tan SH, Tang MBY. Mortality of bullous pemphigoid in Singapore: risk factors and causes of death in 359 patients seen at the National Skin Centre. Br J Dermatol. 2014;170(6):1319-1326. doi: 10.1111/bjd.12806 [DOI] [PubMed] [Google Scholar]
  • 26.Tedbirt B, Andrieu E, Ezzedine K, et al. Mortality in bullous pemphigoid: individual patient data meta-analysis on 12 international studies [in French]. Ann Dermatol Vénéréol. 2019;146(12, suppl):A96. doi: 10.1016/j.annder.2019.09.098 [DOI] [Google Scholar]
  • 27.Taïeb A, Klene C, Maleville J. Immediate treatment of bullous pemphigus with a corticosteroid-cyclophosphamide combination [in French]. Ann Dermatol Venereol. 1986;113(12):1223-1229. [PubMed] [Google Scholar]
  • 28.Devendeville A, Lok C, Buleux E, Denoeux JP, Lienard J. Bullous pemphigoid and its prognostic: retrospective study of 33 patients [in French]. Rev Gériatr. 1991;16(6):249-254. [Google Scholar]
  • 29.Depaire F. Traitement des pemphigoides bulleuses par tétracyclines. Master’s thesis. Université Montpellier I; 1995. [Google Scholar]
  • 30.Wells GA, Shea B, O'Connell D, et al; Ottawa Hospital Research Institute . The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Published 1999. Accessed November 29, 2020. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp
  • 31.World Population Prospects, Population Division, United Nations . Standard projections (estimates and projection variants). Published August 2019. Accessed October 29, 2019. https://population.un.org/wpp/Download/Standard/Mortality/
  • 32.Pflaumer P. Life Table Forecasting with the Gompertz Distribution. American Statistical Association; 2007. doi: 10.17877/DE290R-8212 [DOI] [Google Scholar]
  • 33.Trikalinos TA, Trow P, Schmid CH. Simulation-based comparison of methods for meta-analysis of proportions and rates. Agency for Healthcare Research and Quality; 2013. Accessed November 29, 2020. https://www.ncbi.nlm.nih.gov/books/NBK179162/ [PubMed]
  • 34.Monshi B, Gulz L, Piringer B, et al. Anti-BP180 autoantibody levels at diagnosis correlate with 1-year mortality rates in patients with bullous pemphigoid. J Eur Acad Dermatol Venereol. 2020;34(7):1583-1589. doi: 10.1111/jdv.16363 [DOI] [PubMed] [Google Scholar]
  • 35.Sticherling M, Franke A, Aberer E, et al. An open, multicentre, randomized clinical study in patients with bullous pemphigoid comparing methylprednisolone and azathioprine with methylprednisolone and dapsone. Br J Dermatol. 2017;177(5):1299-1305. doi: 10.1111/bjd.15649 [DOI] [PubMed] [Google Scholar]
  • 36.Kibsgaard L, Rasmussen M, Lamberg A, Deleuran M, Olesen AB, Vestergaard C. Increased frequency of multiple sclerosis among patients with bullous pemphigoid: a population-based cohort study on comorbidities anchored around the diagnosis of bullous pemphigoid. Br J Dermatol. 2017;176(6):1486-1491. doi: 10.1111/bjd.15405 [DOI] [PubMed] [Google Scholar]
  • 37.Bernard P, Venot J, Rommel A, Bonnetblanc JM, Texier L. Bullous pemphigoid: a prognostic study: review of fifty-seven observations [in French]. Sem Hôp Paris. 1986; 62(18):1229-1232. [Google Scholar]
  • 38.Michel JL, Bourdet JF, Perrot JL, et al. Corticosteroid treatment, risk factor of pemphigoid: retrospective analysis of 62 cases [in French]. Nouv Dermatol. 1999;18:331-335. [Google Scholar]
  • 39.Roujeau JC, Lok C, Bastuji-Garin S, Mhalla S, Enginger V, Bernard P. High risk of death in elderly patients with extensive bullous pemphigoid. Arch Dermatol. 1998;134(4):465-469. doi: 10.1001/archderm.134.4.465 [DOI] [PubMed] [Google Scholar]
  • 40.Chevalier V, Barbe C, Reguiai Z, Plée J, Grange F, Bernard P. Impact of neurological diseases on the prognosis of bullous pemphigoid: a retrospective study of 178 patients [in French]. Ann Dermatol Venereol. 2016;143(3):179-186. doi: 10.1016/j.annder.2015.12.016 [DOI] [PubMed] [Google Scholar]
  • 41.Nespoulous L, Dalmay F, Bedane C, Assikar S. Prognostic factors of bullous pemphigoid in 329 patients [in French]. Ann Dermatol Vénéréol. 2018;145(12, suppl):S164-S165. doi: 10.1016/j.annder.2018.09.219 [DOI] [Google Scholar]
  • 42.Cordel N, Renier M, Samyn A, Fauvel C, Hope-Rapp E, Gilbert D; la Société de dermatologie de la Guadeloupe . Epidemiology of bullous pemphigoid in Guadeloupe (French West Indies) [in French]. Ann Dermatol Venereol. 2009;136(12):907-909. doi: 10.1016/j.annder.2009.10.182 [DOI] [PubMed] [Google Scholar]
  • 43.Cantegrit T, Merklen C, Birckel E, et al. One- and three-year mortality in patients with bullous pemphigoid treated as first-line with intensive local corticosteroid therapy [in French]. Ann Dermatol Vénéréol. 2019;146(12, suppl):A161. doi: 10.1016/j.annder.2019.09.215 [DOI] [Google Scholar]
  • 44.Clapé A, Muller C, Plée J, Viguier M, Vanhaecke C, Bernard P. Feasibility and healthcare costs of superpotent topical corticosteroid therapy in bullous pemphigoid: a prospective, observational study in an academic centre in France. Br J Dermatol. 2020;183(4):775-776. doi: 10.1111/bjd.19151 [DOI] [PubMed] [Google Scholar]
  • 45.Serwin AB, Musialkowska E, Piascik M. Incidence and mortality of bullous pemphigoid in north-east Poland (Podlaskie Province), 1999-2012: a retrospective bicentric cohort study. Int J Dermatol. 2014;53(10):e432-e437. doi: 10.1111/ijd.12492 [DOI] [PubMed] [Google Scholar]
  • 46.Kalinska-Bienias A, Lukowska-Smorawska K, Jagielski P, Kowalewski C, Wozniak K. Mortality in bullous pemphigoid and prognostic factors in 1st and 3rd year of follow-up in specialized centre in Poland. Arch Dermatol Res. 2017;309(9):709-719. doi: 10.1007/s00403-017-1772-x [DOI] [PubMed] [Google Scholar]
  • 47.Heilborn JD, Ståhle-Bäckdahl M, Albertioni F, Vassilaki I, Peterson C, Stephansson E. Low-dose oral pulse methotrexate as monotherapy in elderly patients with bullous pemphigoid. J Am Acad Dermatol. 1999;40(5, pt 1):741-749. doi: 10.1016/S0190-9622(99)70156-8 [DOI] [PubMed] [Google Scholar]
  • 48.Kjellman P, Eriksson H, Berg P. A retrospective analysis of patients with bullous pemphigoid treated with methotrexate. Arch Dermatol. 2008;144(5):612-616. doi: 10.1001/archderm.144.5.612 [DOI] [PubMed] [Google Scholar]
  • 49.Fisch A, Morin L, Talme T, Gallais Sérézal I.. Bullous pemphigoid treated with methotrexate: safety profile based on renal function at diagnosis [in French]. Ann Dermatol Vénéréol. 2018;145(12, suppl):S48. doi: 10.1016/j.annder.2018.09.006 [DOI] [Google Scholar]
  • 50.Cortés B, Khelifa E, Clivaz L, et al. Mortality rate in bullous pemphigoid: a retrospective monocentric cohort study. Dermatology. 2012;225(4):320-325. doi: 10.1159/000345625 [DOI] [PubMed] [Google Scholar]
  • 51.Burton JL, Harman RR, Peachey RD, Warin RP. Azathioprine plus prednisone in treatment of pemphigoid. BMJ. 1978;2(6146):1190-1191. doi: 10.1136/bmj.2.6146.1190 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Gudi VS, White MI, Cruickshank N, et al. Annual incidence and mortality of bullous pemphigoid in the Grampian Region of North-east Scotland. Br J Dermatol. 2005;153(2):424-427. doi: 10.1111/j.1365-2133.2005.06662.x [DOI] [PubMed] [Google Scholar]
  • 53.Persson MSM, Harman KE, Vinogradova Y, et al. Incidence, prevalence and mortality of bullous pemphigoid in England 1998-2017: a population-based cohort study. Br J Dermatol. 2020. doi: 10.1111/bjd.19022 [DOI] [PubMed] [Google Scholar]
  • 54.Angit C, King C, Lister KR. Regional audit of bullous pemphigoid. J Am Acad Dermatol. 2011;64(2 suppl 1):AB94. [Google Scholar]
  • 55.Chalmers JR, Wojnarowska F, Kirtschig G, et al. A randomised controlled trial to compare the safety, effectiveness and cost-effectiveness of doxycycline (200 mg/day) with that of oral prednisolone (0.5 mg/kg/day) for initial treatment of bullous pemphigoid: the Bullous Pemphigoid Steroids and Tetracyclines (BLISTER) trial. Health Technol Assess. 2017;21(10):1-90. doi: 10.3310/hta21100 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Mokhtari F, Fariba I, Radfar S, Asilian Mahabadi A, Faghihi G, Hosseini SM. Epidemiological survey of the patients with bullous pemphigoid hospitalized in Alzahra Hospital, Isfahan, Iran, during the years 2006-2015. J Isfahan Med School. 2019;36(507):1491-1497. doi: 10.22122/jims.v36i507.10647 [DOI] [Google Scholar]
  • 57.Kridin K, Shihade W, Bergman R. Mortality in patients with bullous pemphigoid: a retrospective cohort study, systematic review and meta-analysis. Acta Derm Venereol. 2019;99(1):72-77. [DOI] [PubMed] [Google Scholar]
  • 58.Rozenblat M, Halaj A, Rozenblat T, et al. Mortality and risk factors among Israeli bullous pemphigoid patients. Arch Dermatol Res. 2019;311(1):19-27. doi: 10.1007/s00403-018-1875-z [DOI] [PubMed] [Google Scholar]
  • 59.Kanamori S, Nagata T, Uezato H, Nonaka S. Statistical observation of bullous pemphigoid patients reported at the Dermatology Clinic, Ryukyus University Hospital [in Japanese]. Nishi Nihon Hifuka. 2004;66:183-187. doi: 10.2336/nishinihonhifu.66.183 [DOI] [Google Scholar]
  • 60.Sasai S, Asano M, Numata Y, et al. A retrospective statistical analysis of prognostic factors for a lethal outcome in 52 elderly Japanese patients with bullous pemphigoid [in Japanese]. Jap J Dermatol. 2015;125(2):235-241. doi: 10.14924/dermatol.125.235 [DOI] [Google Scholar]
  • 61.Jeon HW, Yun SJ, Lee S-C, Won YH, Lee J-B. Mortality and comorbidity profiles of patients with bullous pemphigoid in Korea. Ann Dermatol. 2018;30(1):13-19. doi: 10.5021/ad.2018.30.1.13 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Nanda A, Al-Saeid K, Al-Sabah H, Dvorak R, Alsaleh QA. Clinicoepidemiological features and course of 43 cases of bullous pemphigoid in Kuwait. Clin Exp Dermatol. 2006;31(3):339-342. doi: 10.1111/j.1365-2230.2005.02040.x [DOI] [PubMed] [Google Scholar]
  • 63.Phoon YW, Fook-Chong SM, Koh HY, Thirumoorthy T, Pang SM, Lee HY. Infectious complications in bullous pemphigoid: an analysis of risk factors. J Am Acad Dermatol. 2015;72(5):834-839. doi: 10.1016/j.jaad.2015.01.029 [DOI] [PubMed] [Google Scholar]
  • 64.Wei L. Clinical analysis of 163 cases of bullous pemphigoid. Master’s thesis. China Medical University; 2018. [Google Scholar]
  • 65.Kulthanan K, Chularojanamontri L, Tuchinda P, Sirikudta W, Pinkaew S. Prevalence and clinical features of Thai patients with bullous pemphigoid. Asian Pac J Allergy Immunol. 2011;29(1):66-72. [PubMed] [Google Scholar]
  • 66.Carvajal Aguilera D, Fernández Moraga J, Valenzuela Ahumada F, Morales Huber C, Fernández Moraga A.. Autoimmune blistering diseases: clinical characterization, therapeutic response and mortality in a university center in Chile [in Spanish]. Piel. 2020;35(1):10-15. doi: 10.1016/j.piel.2019.03.002 [DOI] [Google Scholar]
  • 67.Korman NJ. Bullous pemphigoid: the latest in diagnosis, prognosis, and therapy. Arch Dermatol. 1998;134(9):1137-1141. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement.

eTable 1. Search Strategy in Online Databases

eTable 2. Long-term All-Cause Mortality Assessment

eTable 3. Causes of Death Extracted at 1-Year or Longer Follow-up

eMethods. Data Screening, Extraction, and Analysis

eFigure 1. Forest Plot of All-Cause 1-Year Mortality Rate

eFigure 2. Age Distribution at Inclusion in Pooled Individual Data Studies and Age-Sex Relationship

eFigure 3. Forest Plot of All-Cause Directly Standardized 1-Year Mortality Rates

eFigure 4. Funnel Plot of All-Cause 1-Year Death Rates of the 56 Studies Included in the Meta-analysis

eFigure 5. Assessment of Risk of Bias in the 56 Studies on a Custom Scale

eResults. Comparison of Crude Death Rates, Directly Standardized Death Rates, Causes of Death, and Risk of Bias

Click here for additional data file. (1.2MB, pdf)

Articles from JAMA Dermatology are provided here courtesy of American Medical Association

RESOURCES