Survivorship and Death in Rheumatoid Arthritis.

Survivorship and Death in Rheumatoid Arthritis
Jarnal Uddin, Arthur S. Kraus and H. Garfield Kelly
Data are presented on survivorship and causes of death of 475 patients with
rheumatoid arthritis first seen between 1954 and 1966. The observed cumulative
survival rate was lower than expected i n each sex at each year of follow-up,
particularly in the later years. The adjusted rate was generally lower in males
than in females. Eighteen of the 86 deaths due t o known causes were attributed
to infection, a significantly excessive proportion.
A number of studies are available on the
course and prognosis of patients with rheumatoid arthritis, which deal particularly
with factors affecting functional capacity
and disability.1-6 Little has been written
on longevity and causes of death. Both
Duthie et a17 and VanDam et aP have
compared death rates per thousand patient
years of observation by sex and age without
regard to the trend in mortality, year by
year, following initial contact. Cobb et UP
have graphed cumulative survival rates of
a group of patients with both rheumatoid
arthritis and ankylosing spondylitis from
the time of onset of disease.
This study presents data on survivorship
and causes of death of 475 patients with
probable, definite or classic rheumatoid
arthritis (RA) , according to American
Rheumatism Association criteria,lO from the
time of their entry in the Queen’s University Rheumatic Diseases registry, between
1954 and 1966. Approximately one third of
all patients required no hospital care during this period of time. The analysis of
survivorship takes into consideration not
only death rates per thousand patient years
of observation, but also the cumulative
survival rate from time of entry in the
registry through each successive year of
follow-up during the total period of observation.
METHODS
From the Departments of Medicine, Community
Health and Epidemiology, and the Rheumatic Diseases Unit, Queen’s University, Kingston, Ontario.
JAMAL UDDIN,MD, FRCP (C) : Assistant Resident
in Medicine (Rheumatology) , Kingston Hospital
and Queen’s University. ARTHUR S. KRAUS, ScD,
FAPHA: Associate Professor of Biostatistics, Queen’s
University. H. GARFIELD
KELLY,
MD, CM. FRCP(C),
FACP: Professor of Medicine, Queen’s University.
Reprint requests should be addressed to Dr. H.
Garfield Kelly, Etherington Hall, Queen’s University, Kingston, Ontario, Canada.
Submitted for publication Aug 5, 1969; accepted,
Sept 8, 1969.
To calculate survivorship, it was necessary to
determine the number of patients alive, dead or
lost to follow-up in each successive year of the study
period by correspondence or interview with patients,
relatives, physicians or other interested persons.
Thirty- four patients (7%) were lost to follow-up,
and 94 patients (20%) died during the study period.
Survivorship for all patients was calculated from
data which included sex, date of birth, date of entry
in the rheumatic diseases registry and date of death
or loss from follow-up, by the life-table method
applied by Merrell and Shulman to a survivorship
study of chronic disease in general and systemic
lupus erythematosus (SLE) in particularP
The cumulative survival rate for all patients with
RA was calculated and compared with the expected
survival rate for a general population sample which
Mhritis and Rheumatism, Vol. 13, No. 2 (March-April 1970)
125
UDDIN R A 1
would match the patient group with respect to age
and sex distribution. The expected survival rate
was based on age-sex-specific death rates in Ontario in 1961.'* The statistical significance of differences, when indicated, was determined 'by the method of Greenwood, using the standard error of the
life-table survival rate.I5 An analysis of cause of
death was based on the underlying cause of death
listed on the death certificate and verified by autopsy when available.
RESULTS
Survivorship
T h e observed cumulative survival rate
for RA patients was lower in each sex a t
each year of follow-up than the expected
survival rate for the general population,
with a more pronounced difference in the
later years of follow-up. Through the fifth
year of follow-up in males and the seventh
year of follow-up in females (in which years
there remained 50 person-years of follow-up
in each sex), the difference was statistically
significant (p < 0.01) (Tables 1 and 2,
Fig 1).
T h e relationship between observed and
expected survival rates can be expressed:
adjusted cumulative survival rate
=
observed cumulative survival rate
expected cumulative survival rate
Adjusted survival rates in the early years
of follow-up, which were above 90% in each
sex, gradually fell below BOY0 in later years.
Adjusted survival rates for males were always appreciably lower than those for females, except during the last 2 years of
follow-up when the small numbers remaining in the series could easily have produced,
by chance, an apparent reversal of this relationship (Table 3).
Causes of Death
Of the 94 patients who died, causes of
death were determined from death certificates in 86 and confirmed by autopsy in 28
(Table 4 ) . T h e most frequent cause of
death was cardiovascular disease of all varie126
Table 1. Life-Table Analysis Showing
Observed and Expected Cumulative
Survival Rate in Males With
Rheumatoid Arthritis
Cumulative
Patients
Survival rate (96)
No. of
living at Person
_yr. of
beginning years of
EXfollow up of year* follow-upt Observed pectedz
1
2
3
4
5
6
7
8
9
10
151
108
93
76
60
51
36
26
19
3
135.0
104.0
86.5
71.0
56.0
46.0
32.0
23.5
11.0
1.5
91.9
85.7
81.8
74.8
73.5
65.4
61.3
56.1
56.1
56.1
96.9
94.1
91.0
87.4
83.8
80.0
77.0
74.2
71.5
67.4
* For year 1 = no. entering study group. For
subsequent years = no. alive at the beginning of
previousyear - (deaths
lossto follow up no.
alive when they reached the end of the observation period) during the previous year.
t Person years of follow-up = no. alive at the
beginning of the year - M (no. lost to follow-up
+no. alive when they reached the end of the
observation period) during the year.
3 Based on general population sample.
+
+
ties (51 patients), but the proportionate
mortality for this cause group was lower
than expected. Of special interest was the
high incidence of death from infection (18
patients), which was much higher than the
expected incidence of death from this cause
in the general population sample (Table 5).
Relatively small numbers of patients died
Irom neoplastic disease, renal failure, respiratory insufficiency and gastrointestinal
disease. I n eight instances, the cause of
deatli could not be determined.
DISCUSSION
Three previous studies have documented
decreased survival of patients with
T h e first of these by Cobb et U P presented
Arthritis and Rheumatism, Vol. 13, No. 2 (March-April 1970)
SURVIVORSHIP AND DEATH IN RA
Table 2. Life-Table Analysis Showing
Observed and Expected Cumulative
Survival Rate in Females With RA
Table 3.
Cum u Iat ive
Survival Rate (I)
Patients
No. of
living at Person
EXyr. of beginning years of
follow up of year* follow-uptobserved pectedl
1
2
3
4
5
6
7
8
9
10
324
260
217
175
146
113
88
56
41
26
299.0
241.5
199.0
164.5
132.5
102.0
74.5
49.5
35.5
13.5
95.3
92.9
90.1
85.7
81.8
79.4
74.1
71.1
63.1
58.4
97.9
95.6
93.4
91.2
89.5
87.3
84.0
82.1
80.2
78.5
q
+
4
0
.
,
2
. . .
.
.
. . .
4
6
8
1
YEARS Of FOLLOW-UP
1
2
3
4
5
6
7
8
9
10
Males
Female
94.8
91.1
89.9
85.6
87.7
81.8
79.6
75.6
78.5
83.2
99.9
97.2
96.5
94.0
91.4
91.0
88.2
86.6
78.7
74.4
observed cumulative survival rate
expected cumulative survival rate
-
40
No. of yr.
of follow-up
* Adjusted cumulative survival rate =
* For year 1 = no. entering the study group; for
subsequent years = no. alive at the beginning of
loss to follow-up
previous year - (deaths
no. alive when they reached the end of the observation period) during the previous year.
t Person years of follow-up = no. alive at the
beginning of the year - % (no. lost to follow-up
+no. alive when they reached the end of the
observation period) during the year.
$ Based on general population sample.
+
Adjusted Cumulative Survival Rate*
(96) in Males and Females With RA
0
Fig 1. Observed and expected cumulative surviva1 rate in males and females with RA.
Arthritis and Rheumatism, Vol. 13, No. 2 (March-April 1970)
.
Table 4. Cause of Death of 94 RA Patients
Cardiovascular
Cardiac
Cerebrovascular
Pulmonary emboli &/or infarction
Mesenteric th rombosis
Ruptured abdominal aneurysm
Postoperative shock
Total
Infectious
Pulmonary
Septicemia and pyemia
Postgastrectorny peritonitis
Total
Respiratory
(Chronic Pulmonary Disease)
Neoplastic
Renal
Amyloidosis
Necrotizing papillitis
HeDatorenal failure
Total
Gastrointestinal
GI hemorrhage
Chronic pancreatitis
Total
Unknown
Total all causes
4 ( 3)1
1
2 ( 01-
a ( 0)
94 (28)
* Number in parentheses indicates no. confirmed by autopsy.
127
UDDlN E l AL
Table 5.
Comparison of Observed Leading Causes of Death of Male and Female
RA Patients With Those Expected in a General Population of the Same Age-Group
Male
Cause of death
Cardiac
Infectious
Throm boem bol ic
Neoplastic
Female
% Observed
% Expected
% Observed
% Expected
35.30
26.47
8.82
8.82
48.61
2.62
12.09
12.39
38.50
19.23
26.92
7.69
41.65
2.53
20.60
6.36
data on 583 patients with RA and ankylosing spondylitis who were hospitalized and
subsequently followed for an average period
of 9.6 years. Observed death rates for males
and females under 50 years of age were, respectively, 20.6 and 10.7/ 1000 patient years
of observation, in contrast to expected death
rates for males and females of 3.9 and 2.9,
respectively. Over the age of 50, observed
and expected death rates did not differ
significantly for either sex. The higher
mortality for younger males could not be
correlated with the presence or absence of
spondylitis.
Duthie et al? have reported findings on
the course and prognosis of 307 patients
with RA severe enough to require hospital
admission. At the time of final assessment,
after an average follow-up period of 9 years,
75 patients were dead and 32 were lost to
follow-up. Death rates per thousand years
of observation by age and sex revealed the
observed death rate to be higher than the
expected rate in all three age groups: under
50, from 50-59, and over 60 in both sexes.
The report by VanDam et U P was based
on a follow-up of 1214 patients with R A
over a period of 11 years, during which period of time 229 patients died. The observed
death rate was approximately 60% higher
than the expected rate in both sexes. When
the disease began before 50 years of age, the
adjusted death rate was about 50% higher
than when it began after the age of 50.
Our data indicate decreased survival of
RA patients in all age and sex groups
(Table 6 ) . It might appear that the recent
study by Burch14 on Pima Indians is at
variance with this finding and those previously cited. Although Burch did not find
a higher mortality among Indians with
probable or definite RA than “in all others,”
his “all others” classification included an
appreciable number of patients with a positive test for rheumatoid factor who. were
shown to have an excessive mortality rate.
Therefore, Burch’s data are not comparable
with and do not contradict those of the
previously noted authors.
Our data also demonstrate that in a given
population of patients with RA of all
grades of severity, the difference between
observed and expected survival rates becomes statistically significant during later
years of follow-up in both sexes (Table 1
and 2, Fig 1). Adjusted survival rates for
males were almost always appreciably below those for females.
As noted in Fig 1, the trend of the observed cumulative survival rate for males is
essentially linear in its downward course
during the first 8 years; whereas, for females,
it appears to decrease less sharply in the
first few years of follow-up than later on.
This trend is in contrast to that of many
other chronic diseases, particularly those
that are obviously life-threatening. For example, in SLE15-17 and cancer of the
128
Arthritis and Rheumatism, Vol. 13, No. 2 (March-April 1970)
SURVIVORSHIP AND DEATH
IN RA
breast,ls the decrease in survival was rela- death, or the type and duration of treattively greater in the first few years of follow- ment. The majority of patients who died
from infection were receiving steroid, but
UP.
The distribution of causes of death in so were those who died from cardiac,
our group of patients did not differ signifi- thromboembolic and neoplastic disorders.
cantly from those in other reports, all of It is difficult to evaluate the specific role of
which show a relatively high incidence of steroid in increasing the incidence of inEecdeath from infection (Table 7). In our tion in the presence of so many compliseries, the observed incidence of death from cating factors-the debilitating nature of
infection was about ten times higher than the disease, the presence of coexisting disexpected in males and about seven times ease, and the possible adverse effects of the
higher than that expected in females, both other forms of treatment.
differences being statistically significant (p
T h e infectious cause group was the only
< .001) (Table 5).
one which showed a pronounced relative
It has not been possible to correlate the excess (Table 5 ) . However, only 18 deaths
incidence of infection with the duration were attributed to infection; whereas, total
and severity of disease, the mean age of observed deaths (94) exceeded total ex-
Table 6.
Observed and Expected Death Rates per 1000 Patient Years of Observation
by Age and Sex
Male
Age (yrs)
< 50
5059
60 and >
Observed Expected*
Patient yrs death rate death rate Patient yrs
159.0
138.5
272.0
* Based on general
Table 7.
Female
6.3
43.3
117.6
3.8
13.3
63.6
389.5
304.0
622.0
Observed Expected*
death rate death rate
10.3
19.7
72.3
2.1
8.9
43.2
population sample.
Comparison of Causes of Death of RA Patients in Different Series Studies
VanDam
Cobb
Cause of Death
Distribution
Card iovascutar*
Infection
Cancer
Renal disease
Gastrointestinal disease
Respiratory disease
Miscellaneous
Unknown
Uddin
Duthie
137 patients
229 patients 94 patients
75 patients
(20)
(28)
(4011
35.0
23.4
10.9
12.4
5.8
45.3
14.7
13.3
17.3
8.0
38.5
12.6
15.2
7.8
6.5
7.3
5.1
1.3
16.9
2.2
-
-
-
-
(96)
51.0
19.1
7.4
4.2
2.1
4.2
3.2
8.5
* Includes all deaths from heart and vascular diseases.
t Number in parentheses indicates no. of autopsies.
Arthritis and Rheumatism, Vol. 13, No. 2 (March-April 1970)
123
UDDIN ET A 1
pected deaths (60) by 34, making it appare n t that causes other than infection also
contributed to increased mortality.
Hopefully, therapeutic measures will soon
become available which will increase longevity a n d narrow the g a p between the
observed mortality rate for RA patients and
the expected mortality rate for the general
population.
1961. Vol. 1. Relazioni. Minerva Medica.
Torino, Italy, 1961, p. 161.
9. COUB,S., ANDERSON,
F. and BAUER,W.
Length of life and cause of death in rheumatoid arthritis. New Eng J Med 249:553, 1953.
G. A., Conu, S.,
10. ROPES,M. W., BENNETT,
JACOX,R. and JESSAR,R. A. 1958 revision of
diagnostic criteria for rheumatoid arthritis. Bull
Rheum Dis 9:175, 1958.
11. MERRELL,
M. and SHULMAN,
L. E. Determination of prognosis in chronic disease,
illustrated by systemic lupus erythematosus.
REFERENCES
J Chron Dis 1:12, 1955.
I. RACAN,C. and FARRINGTON,
E. The
12. Vital Statistics, 1961. Dominion Bureau
clinical features of rheumatoid arthritis. Prog- of Statistics, Health and Welfare Division, Vital
nostic indices. JAMA 181:663, 1962.
Statistics Section, Ottawa, Canada.
2. SHARP,J. T., CALKINS,
E., COHEN,A. S.,
13. GREENWOOD,
M. A report on the natural
SCHUBART,
A. F. and CALABRO,
J. J. Observa- duration of cancer. Appendix I. The errors of
tions on the clinical, chemical and serological sampling of the survivorship tables. Reports on
manifestations of rheumatoid arthritis, based Public Health and StatistiGal Subjects No. 33.
on the course of 154 cases. Medicine 43:41, 1964. London, 1926.
3. SHORT,C. L. Rheumatoid arthritis: Types
14. BURCH,T. A. I n discussion on rheumaof course and prognosis. Med Clin N Amer toid factor in the Blackfeet and Pima Indians.
52:549, 1968.
Population studies of the rheumatic diseases.
4. BYWATERS,
E. G. L. and DRESNER,
E. The P.H. International Congress series No. 148. Exprognosis in rheumatoid arthritis. Quart J Med cerpta Medica Foundation, July 1968. p. 200.
213463, 1952.
15. SIECEL,M., GWON,N., LEE, S. L., RIVERO,
5. BROWN,
P. E. and DUTHIE,J. J. R. Varia- I., and WoNc, W. Survivorship in systemic
tions in the course of rheumatoid arthitis. A n n lupus erythematosus: relationship to race and
Rheum Dis 17359, 1958.
pregnancy. Arthritis Rheum I2:117, 1969.
6. SIEVERS,K. T h e rheumatoid factor in
16. KELLUM,R. E. and HASERICK,
J. R.
definite rheumatoid arthritis. An analysis of Systemic lupus erythematosus. A statistical
1279 adult patients with a follow-up study. Acta evaluation of mortality based on consecutive
Rheum Scand (Suppl):9, 1965.
series of 299 patients. Arch Int Med I13:200,
1964.
7. DUTHIE,J. J. R., BROWN,P. E., TRUELOVE,
L. H., BARACAR,
F. D. and LAWRIE,A. J. Course
T. Long term prognosis of
17. LEONHARDT,
and prognosis in rheumatoid arthritis. A fur- systemic lupus erythematosus. Acta Med Scand
ther report. Ann Rheum Dis 23:193, 1964.
445 (Suppl):440, 1966.
G., LEZWIJN,
A. and Bos, J. G.
8. VANDAM,
P., GAL18. SMITHERS,
D. W., RICBY-JONES,
Death-rate of patients with rheumatoid arthritis. TON,D. A. G. and PAYNE,
P. M. Cancer of the
Atti del X Congress0 della lega Internazionale breast. A review. Brit J Radiol 4 (Suppl):51,
Contro il Rheumatismo, Roma, 3-7 September, 1952.
130
Arthritis and Rheumatism, Vol. 13, No. 2 (March-April 1970)