Funding sources None.
Epidemiology and Health Services Research
Mortality of bullous pemphigoid in Singapore: risk factors and causes of death in 359 patients seen at the National Skin Centre†
Article first published online: 29 APR 2014
© 2013 British Association of Dermatologists
British Journal of Dermatology
Volume 170, Issue 6, pages 1319–1326, June 2014
How to Cite
Cai, S.C.S., Allen, J.C., Lim, Y.L., Chua, S.H., Tan, S.H. and Tang, M.B.Y. (2014), Mortality of bullous pemphigoid in Singapore: risk factors and causes of death in 359 patients seen at the National Skin Centre. British Journal of Dermatology, 170: 1319–1326. doi: 10.1111/bjd.12806
Conflicts of interest None declared.
Plain language summary available online.
- Issue published online: 19 JUN 2014
- Article first published online: 29 APR 2014
- Accepted manuscript online: 24 DEC 2013 08:08AM EST
- Manuscript Accepted: 18 DEC 2013
Bullous pemphigoid (BP) is the most common autoimmune-mediated subepidermal blistering skin disease and is associated with significant morbidity and mortality.
To determine the 3-year mortality rate, risk factors and causes of death in patients with BP in Singapore, compared with the general population.
We conducted a retrospective cohort study of all newly diagnosed patients with BP seen at the National Skin Centre from 1 April 2004 to 31 December 2009. Demographic and clinical data on comorbidities and treatment were recorded. Mortality information was obtained from the National Registry of Diseases.
In total 359 patients were included in our study. The 1-, 2-, 3-year mortality rates were 26·7%, 38·4% and 45·7%, respectively. The 3-year standardized mortality risk for patients with BP was 2·74 (95% confidence interval 2·34–3·19) times higher than for the age- and sex-matched general population. Parkinson disease, heart failure and chronic renal disease were associated with increased mortality, while combination treatment with low-to-moderate-dose corticoste-roids and immunomodulatory agents such as doxycycline and/or nicotinamide was associated with lower mortality. Overall, infections were the most common cause of death (59·8%), with the main causes of death being pneumonia (42·7%), cardiovascular disease (14·6%) and stroke (11·6%).
This study confirms an increased 3-year mortality rate for patients with BP in Singapore. Risk factors for increased mortality include medical comorbidities, especially neurological, cardiac and renal diseases. Treatment with combination therapy, including the use of low-to-moderate-dose corticosteroid, appeared to decrease mortality risk in patients with BP.
Bullous pemphigoid (BP) is an autoimmune disorder that causes immune-mediated blistering of the skin and mucous membranes. BP is the most common immunobullous disorder in Singapore, with an estimated incidence of 7·6 per million per year and a predilection for elderly Chinese patients, with an average age of onset of 77 years. With the rapidly ageing population in Singapore, it is expected that the incidence of BP will increase in tandem. It constitutes 88% of all subepidermal immunobullous disorders seen at the National Skin Centre, which is the major referral centre for immunobullous skin diseases in Singapore.
BP has a chronic relapsing course with high morbidity and mortality, with reported first-year mortality rates of 6–44% in Western cohorts.[2-17] Patients also often require long-term immunosuppressive therapy with systemic corticosteroids and other agents, which may further increase the risk of morbidity and mortality.[7-12]
Presently, there are limited data on mortality rates of BP in Asian patients. Additionally, it is unknown whether Asian patients with BP have different mortality risk factors from Western cohorts. Our primary aim was to determine the 3-year mortality rate of patients with BP in Singapore and to compare it with the rate in the general population. Our hypothesis was that the 3-year mortality rate of patients with BP would be higher than that in the age- and sex-matched general population. Secondary aims included identifying the respective causes of death and prognostic factors affecting survival during the 3-year follow-up.
Patients and methods
This was a retrospective cohort study. Our study cohort comprised all newly diagnosed Singaporean patients with BP seen at the National Skin Centre from 1 April 2004 to 31 December 2009. A diagnosis of BP was based on the presence of at least three of the four following criteria: (i) clinical findings consistent with BP, e.g. tense blisters arising from erythematous or urticated lesional skin affecting the trunk and limbs; (ii) histopathological findings of a subepidermal blister; (iii) direct immunofluorescence findings of linear deposition of IgG and/or complement along the dermoepidermal junction; and (iv) indirect immunofluorescence findings of either roof pattern or roof and floor pattern, or positive anti-BP180 NC16A IgG antibodies measured by enzyme-linked immunosorbent assay (Medical and Biological Laboratories Co. Ltd, Nagoya, Japan). Exclusion criteria included foreign patients and patients with any other immunobullous disorder such as pemphigus, linear IgA bullous dermatosis and bullous lichen planus.
Patients were identified through the National Skin Centre's electronic medical and histology records. Data collection was performed by reviewing the case records of the patients to obtain demographic data, medical history, ambulatory status, clinical features, laboratory results, treatment details and disease status at the last follow-up visit. Patients were then matched with the National Registry of Diseases Office (NRDO) death registry for verification of mortality status and cause of mortality.
The primary aim was to investigate survival of patients with BP during the 3 years following diagnosis. Mortality rates at 1, 2 and 3 years following the diagnosis of BP were calculated using the Kaplan–Meier product limit estimator. Age- and sex-adjusted mortality rates of patients in the BP cohort were calculated using age- and sex-specific life tables provided by the Singapore Department of Statistics, and compared statistically with the corresponding age- and sex-adjusted mortality rates for the general population in Singapore using standardized mortality ratio (SMR) methodology. SMRs were compared for the entire cohort and for age and sex categories. The hypothesis test H0: SMR = 1, H1: SMR > 1 was tested at the 0·05 significance level using 95% confidence intervals (CIs) calculated on the SMR. An SMR of 1 means that observed and expected age- and sex-adjusted mortality rates are equal. Cause-specific SMRs were also calculated based on age- and sex-specific mortality rates for specific causes of death (if available) for the general population.
To assess the association between a particular prognostic factor and mortality rate, the log-rank test in conjunction with Kaplan–Meier analysis, and univariate and multivariate Cox regression analyses were used. Potential prognostic factors included sex, age at diagnosis, ethnicity, clinical features, ambulatory status, disease status at last follow-up, comorbidities, systemic treatment and presence of autoantibodies. Statistical significance was set at P ≤ 0·05. All analyses were performed using SAS 9.2 (SAS Institute Inc., Cary, NC, U.S.A.).
Characteristics of the study cohort
In total 359 patients with BP were included in the study. Of these patients, 187 (52·1%) were female and 172 (47·9%) were male (Table 1). The mean age ± SD at which BP was diagnosed was 75·7 ± 2·6 years and the median age was 77 years (range 27–100). The mean duration of clinical follow-up at the National Skin Centre was 2·1 years, but all patients had complete verification of death status at 3 years using data from the NRDO death registry. The ethnic distribution comprised 300 (83·6%) Chinese, 27 (7·5%) Malay and 32 (8·9%) Indian and other ethnicities. The majority of patients had generalized BP (75·5%).
|Risk factor||Risk factor groups||Univariate analysis (Kaplan–Meier)||Multivariate analysis (Cox regression)a,b|
|Patients, %||Mortality, % (n/N)||P-valuec||HR (95% CI)||P-valued||HR (95% CI)||P-valuee|
|Male||47·9||46·5 (80/172)||0·52||1·11 (0·86–1·50)||0·52||–||–|
|> 77||49·6||60·1 (107/178)||< 0·001||2·19 (1·59–3·03)||< 0·001||1·23 (0·81–1·85)||0·34|
|≤ 77||50·4||31·5 (57/181)|
|Chinese||83·6||46·7 (140/300)||0·02||2·26 (1·11–4·61)||0·03||1·71 (0·76–3·86)||0·20|
|Malay||7·5||59 (16/27)||0·005||3·37 (1·44–7·88)||0·005||2·16 (0·83–5·61)||0·11|
|Indian or other||8·9||25 (8/32)|
|Localized||24·2||47 (41/87)||0·86||1·03 (0·72–1·47)||0·86||–||–|
|Present||5·3||21 (4/19)||0·04||0·37 (0·14–0·99)||0·05||0·43 (0·13–1·36)||0·15|
|Disease status at last follow-up|
|Active or tapering disease||44·8||67·1 (108/161)||< 0·001||4·32 (3·02–6·16)||< 0·001||4·40 (2·94–6·58)||< 0·001|
|Premorbid status at first visit|
|Wheelchair/bedbound||49·6||61·2 (109/178)||< 0·001||2·70 (1·90–3·85)||< 0·001||1·67 (1·08–2·56)||0·02|
|Combinationf||37·6||37·0 (50/135)||< 0·001||0·32 (0·18–0·55)||< 0·001||0·27 (0·14–0·52)||< 0·001|
|Corticosteroids only||50·4||49·2 (89/181)||0·01||0·48 (0·29–0·80)||0·01||0·47 (0·26–0·86)||0·01|
|Other IMAs only||6·4||74 (17/23)|
|Initial corticosteroid dose|
|> 25 mg daily||53·2||41·9 (80/191)||0·09||0·77 (0·57–1·05)||0·09||–||–|
|≤ 25 mg daily||46·8||50·0 (84/168)|
|Discharge corticosteroid dose|
|> 30 mg daily||22·8||44 (36/82)||0·83||0·96 (0·66–1·39)||0·83||–||–|
|≤ 30 mg daily||77·2||46·2 (128/277)|
|Ischaemic heart disease|
|Yes||22·0||65 (51/79)||< 0·001||1·84 (1·32–2·56)||< 0·001||1·23 (0·82–1·85)||0·33|
|Yes||5·6||80 (16/20)||< 0·001||2·74 (1·63–4·61)||< 0·001||2·09 (1·13–3·82)||0·02|
|Yes||59·3||50·2 (107/213)||0·04||1·40 (1·01–1·93)||0·04||0·97 (0·67–1·40)||0·86|
|Yes||32·6||46·2 (54/117)||0·86||1·03 (0·74–1·43)||0·86||–||–|
|Yes||4·2||60 (9/15)||0·29||1·43 (0·73–2·80)||0·30||–||–|
|Yes||11·4||59 (24/41)||0·19||1·34 (0·87–2·06)||0·19||–||–|
|Yes||2·8||60 (6/10)||0·31||1·52 (0·67–3·42)||0·32||–||–|
|Chronic lung disease|
|Yes||6·4||57 (13/23)||0·41||1·27 (0·72–2·23)||0·41||–||–|
|Chronic renal disease|
|Yes||11·4||7 (28/41)||< 0·001||2·00 (1·33–3·00)||< 0·001||1·81 (1·11–2·95)||0·02|
|Yes||40·4||63·4 (92/145)||< 0·001||2·32 (1·70–3·16)||< 0·001||1·38 (0·95–1·99)||0·09|
|Yes||23·7||60 (51/85)||0·001||1·71 (1·23–2·38)||0·002||1·31 (0·88–1·97)||0·19|
|Yes||10·3||70 (26/37)||< 0·001||2·05 (1·34–3·11)||< 0·001||1·85 (1·15–2·96)||0·01|
|Yes||13·4||65 (31/48)||0·005||1·75 (1·18–2·59)||0·005||1·19 (0·76–1·85)||0·45|
|Yes||88·6||47·2 (150/318)||0·10||3·03 (0·75–12·2)||0·12||–||–|
Systemic treatments included corticosteroids (88·0%), doxycycline and/or nicotinamide (25·9%), dapsone (13·9%) and azathioprine (3·9%), including 37·6% who received combination therapy of corticosteroids and immunomodulatory agents. Of those receiving systemic treatment, the majority (65·2%) were treated with a low-to-moderate dose of corticosteroid at ≤ 0·5 mg kg−1 daily. The mean initial dose of systemic corticosteroid used was 0·49 mg kg−1 daily. There were 19 patients with BP who were treated with topical mid-to-high-potency corticosteroids alone, due to limited disease. The distribution of comorbidities is presented in Table 1.
Mortality and standardized mortality ratio
The 1-year mortality was 26·7% (95% CI 22·2–31·6), 2-year mortality 38·4% (95% CI 33·4–43·7) and 3-year mortality 45·7% (95% CI 40·4–51·0; Fig. 1). The calculated age- and sex-adjusted 3-year SMR was significantly elevated for all age groups (Table 2). The SMR was 2·74 (95% CI 2·34–3·19) for the entire BP cohort, and diminished with age as shown in Table 2.
|Age group||Total n||Observed deaths||95% CIa||Expected deaths||SMR||95% CIa|
|Age < 70 years|
|Age 70–79 years|
|Age 80–89 years|
|Age ≥ 90 years|
Risk factors for mortality
As shown in Table 1, age > 77 years (median cohort age) and ethnicity had a significant impact on the risk of death in the univariate analysis, but not in the multivariate analysis.
Ambulatory status and disease status at the last follow-up visit had a significant association with mortality rate. Patients with BP who were wheelchair or bedbound [hazard ratio (HR) 1·67, 95% CI 1·08–2·56] at diagnosis had significantly higher mortality than those who were ambulant (Table 1). Patients who continued to have active BP disease at the last follow-up visit had significantly increased mortality compared with those in remission (HR 4·40, 95% CI 2·94–6·58).
The cardiovascular diseases investigated were heart failure and ischaemic heart disease. Of the two, heart failure was significantly associated with higher mortality (HR 2·09, 95% CI 1·13–3·82; Table 1).
For neurological diseases, stroke, dementia and Parkinson disease were significantly associated with increased mortality in the univariate analysis. However, only Parkinson disease remained significantly associated with elevated mortality in the multivariate analysis (HR 1·85, 95% CI 1·15–2·96).
Chronic renal disease was associated with higher mortality (HR 1·81, 95% CI 1·11–2·95) even after adjusting for other covariates.
Patients with BP who received combination therapy of a low-to-moderate dose of corticosteroids and alternative agents, such as doxycycline and/or nicotinamide (64·4%), had significantly reduced mortality compared with patients who received these alternative agents alone, with an associated HR of 0·27 (95% CI 0·14–0·52; Table 1). Patients treated with a low-to-moderate dose of corticosteroid monotherapy also showed reduced mortality (HR 0·47, 95% CI 0·26–0·86) compared with those prescribed doxycycline and/or nicotinamide agents alone. Higher initial systemic corticosteroid dose trended towards correlation with decreased mortality in the univariate analysis (log-rank test P = 0·09), when the median dose of prednisolone 25 mg daily was used as the cut-off.
Specific causes of death
Overall 164 patients (45·7%) died during the 3 years following the diagnosis of BP (Table 3). Infectious diseases were the most common cause of death (59·8%), specifically due to pneumonia (42·7%), urinary tract infection/pyelonephritis (8·5%), skin and soft tissue infection (3·0%) and septicaemia (3·0%). Other major causes of death included cardiovascular diseases (14·6%), including myocardial infarction (5·5%) and congestive heart failure (4·3%); cerebrovascular accident (11·6%); malignancy (5·5%) and chronic renal disease (4·3%).
|Cause of death||n||%|
|Urinary tract infection or pyelonephritis||14||8·5|
|Skin and soft tissue disease||5||3·0|
|Hepatobiliary sepsis or cholangitis||2||1·2|
|Ischaemic heart disease||5||3·0|
|Hypertensive heart disease||3||1·8|
|Congestive heart failure or coronary insufficiency||7||4·3|
|Ischaemic bowel disease||1||0·6|
|Chronic constipation with toxic megacolon||1||0·6|
|Chronic renal disease||7||4·3|
|Chronic obstructive lung disease||1||0·6|
|Toxic epidermolysis necrosis||1||0·6|
When compared with cause-specific mortality rates for the general population, the age- and sex-adjusted 3-year SMRs were significantly elevated for heart failure (SMR 68·17, 95% CI 27·41–140·46, not shown). However, SMRs were not significantly increased for pneumonia, septicaemia and malignancy. We were unable to compute the SMRs for the other causes of death because specific mortality rates for these causes in the general population were not available.
In this study of 359 Asian patients with BP from the National Skin Centre, we observed 1-, 2- and 3-year mortality rates of 26·7%, 38·4% and 45·7%, respectively. Most significantly, this study found a twofold increase in the 3-year mortality rate for BP in Singapore compared with the general population after adjusting for age and sex, confirming our hypothesis. The SMR in the entire cohort was 2·74, consistent with previous Western cohorts.[2-6, 14] The SMR was increased across all age groups but diminished with increasing age, as also reported by Cortés et al.
Compared with studies in Europe and the U.S.A., which reported higher 1-year mortality rates of up to 44%,[2-10, 16] the 1-year mortality rate in this study was lower at 26·7%. Interestingly, a recent study from China also reported a relatively low 1-year mortality rate of 12·9%. These results suggest there may be differences in mortality risks between white and Asian patients with BP in their first year following BP diagnosis.
There are many conflicting reports on whether old age is an independent risk factor for mortality in BP.[3, 8-10, 12-17] Although the median age of 77 years in our cohort was comparable with that in other studies,[2-11, 13-17] we found that older age was not a significant predictive factor of mortality after multivariate analysis, even though it was significant in the univariate analysis. This lack of association is likely attributable to the marked association between older age and an increased number of underlying comorbidities that contribute to mortality risk.
We found no mortality difference based on sex,[3, 8-10, 13-15] ethnicity, lesion distribution,[8, 12, 14, 15] presence of mucosal lesions,[3, 8, 9, 12-15] positive autoantibodies[3, 12-16] or malignancy[3, 8-10, 12, 16] as reported by other studies. These findings suggest that inherent factors such as sex and race, as well as disease severity at the time of diagnosis, may not influence mortality as much as concomitant comorbidities. However, patients who continued to have active BP disease at the last follow-up visit were found to have an increased risk of death. Cortés et al. also reported that the majority of their study patients with known disease activity had ongoing disease and treatment at the time of death. Poor mobility at time of diagnosis of BP, specifically being wheelchair- or bedbound, was also associated with higher mortality compared with ambulant status. Poor general condition or being bedridden is consistently a predictive factor of death.[8, 14-16, 19] The main causes of death in the study patients with poor mobility included pneumonia (43·1%), cerebrovascular accident (14·7%) and urinary tract infection (7·3%). The higher predisposition to infections may have resulted from physical dependency. It may be worth investigating whether interventions such as physiotherapy, to improve patients' mobility status, may help to reduce mortality in BP.
Of note, we found that cardiac disease, particularly heart failure, significantly increased mortality. The age- and sex-adjusted SMR for death due to heart failure among patients with BP was dramatically elevated compared with the general population, which was not reported by previous studies. While the pathomechanism is unclear, diuretics, which are commonly used in heart failure, have been found to be a risk factor for the development and aggravation of BP in the elderly.[20-22] It is uncertain whether the use of diuretics contributed to a higher mortality risk. This association between heart failure and increased mortality clearly warrants further investigation.
Various studies have established that patients with neurological conditions including stroke, dementia and Parkinson disease have increased prevalence of BP.[22-26] It is noteworthy that concomitant neurological disease at diagnosis was significantly predictive of increased mortality in our study, which is consistent with other studies.[3, 10, 12] This effect was significant even after accounting for poor mobility, suggesting that neurological diseases in themselves contribute to this poor prognosis in patients with BP.
Most interestingly, we found Parkinson disease to be strongly associated with mortality, which also has not been reported previously. The pathogenic mechanism underlying the association between neurological disease and BP is not yet fully understood, but an autoimmunity against hidden BP antigens in the brain,[27-29] via a compromised blood–brain barrier, that cross-react with the skin have been postulated as a possible trigger. Patients with Parkinson disease are known to have dysfunctional blood–brain barriers that may facilitate the crossing of BP autoantibodies into the central nervous system, worsening complement-mediated neuroinflammation and outcome.[31-36]
Another interesting finding was the strong association between chronic renal disease and decreased survival, which other studies have not shown.[10, 12] Notably, our study cohort had significantly more patients with chronic renal disease (11·4%) than the studies by Parker et al. (3·6%) and Li et al. (5·7%). We postulate that this may be attributable to higher rates of renal disease in Asian diabetic patients, as 48·8% of our patients with chronic renal disease had concomitant diabetes mellitus.
In addition, out of seven patients who died from chronic renal disease, six had diabetes mellitus and one who died from diabetes mellitus also had diabetic nephropathy as the secondary cause of death. In total 33% of our patients had diabetes at the time of diagnosis of BP, compared with only 6–15% in other studies.[3, 8, 10-12, 16]
Notably, another significant finding in our study was that treatment with low-to-moderate-dose corticosteroids either alone or in combination with alternative agents such as doxycycline and/or nicotinamide was associated with lower mortality, compared with using alternative noncorticosteroid agents alone. The associations stayed significant even after adjusting for older age, poor mobility, active disease status at last follow-up visit and comorbidities. This is in contrast to the study by Rzany et al., where no significant difference in first-year mortality was found among patients with BP who took combination therapy, corticosteroids alone or tetracyclines alone. Their patients with BP were given a higher corticosteroid dosage on discharge, which is associated with greater mortality and adverse effects.[9, 38] In contrast, our study patients used a lower corticosteroid dose, with a median starting dose of prednisolone 30 mg daily or 0·5 mg kg−1 daily, as per our hospital's treatment guidelines. We believe the judicious use of systemic corticosteroids can improve survival by controlling the BP disease while minimizing the corticosteroid side-effects. This is consistent with studies by Joly et al.[8, 39] that found no significant differences in 1-year survival and complication rates in patients with moderately severe BP treated with topical corticosteroids or daily prednisone 0·5 mg kg−1.
Further survival benefit may have been conferred by the combination therapy of corticosteroids and these steroid-sparing adjuvant agents, enabling physicians to lower the systemic corticosteroid dose. Indeed, this is in agreement with a previous study that showed that tetracycline plus nicotinamide may be a useful alternative steroid-sparing regimen in the treatment of BP. Moreover, corticosteroid therapy alone or in combination was not associated with increased sepsis-related mortality in our study. Corticosteroid side-effects such as peptic ulceration and diabetes were not reported among the major causes of death.
The most frequent causes of death in our BP cohort were infectious diseases (primarily pneumonia), cardiovascular diseases and stroke, which is consistent with previous findings.[3, 8, 15, 16] Of note, in our cohort, the rate of death due to pneumonia or sepsis was not significantly higher than that expected in the national population. A study by Langan et al. found an increased risk of acute pneumonia but not of sepsis after the diagnosis of BP. In contrast, other studies[3, 16] reported a higher rate of sepsis-related death among patients with BP. However, death due to pneumonia was not assessed specifically in these studies.
Strengths of this study include the large cohort, comprehensive patient capture at a national dermatology referral centre, complete verification of mortality status and causes of death with the national registry of deaths, detailed recorded information and long-term follow-up for mortality at 3 years. However, given the retrospective nature of the study, specific effects of disease severity or certain combinations of immunosuppressive agents on mortality could not be evaluated. In addition, it would have been better to compare the mortality rate with that of subjects matched for comorbidities, treatment and other risk factors rather than comparing with the general population, but these data were not available. BP severity was not assessed as this was a retrospective study and there was no published validated BP severity scoring system at the time of study initiation. While treatment protocols were standardized as far as possible, individual treatment variation was inevitable in some cases, which is expected in clinical practice.
In summary, this is a landmark long-term study of Asian patients with BP in a large cohort, which found an increased 3-year mortality rate for BP compared with the general population, with an SMR of 2·74 and a mortality rate of 45·7%. We have found additional risk factors for increased mortality including concomitant comorbidities, especially neurological, cardiac and renal diseases, which will enable better patient prognostication. Treatment with systemic combination therapy, including low-to-moderate doses of corticosteroids, appeared to decrease mortality risk in patients with BP. Our findings are highly relevant for clinical practice and may guide us towards reducing associated mortality, and better prognostication and management of patients with BP.
We would like to thank our colleagues and nurses, especially those in the immunobullous clinic, for their expert care in managing our patients. We also thank the team from the NRDO, comprising Dr Chow Khuan Yew, Ms Sandra Lim, Ms Wong Ching Yee, Ms Janice Fung, Ms Patricia Tan, Ms Lim Gek Hsiang and Mr William Ho, for their invaluable help and support in providing the mortality data for this study, and enabling us to conduct the data analysis at the NRDO.
- 18Department of Statistics Singapore. Yearbook of Statistics Singapore 2011. Singapore: Ministry of Trade and Industry, 2011.