- Email: [email protected]
Lens changes and the incidence of cardiovascular events among persons with diabetes
with two-dimensional Doppler echocardiographic measurement of aortic valve area. Circulation 1983;68:998. Masuyama T, Kodama K, Kitabatake A, et al. Noninvasive evaluation of aortic regurgitation by continuous-wave Doppler echocardiography. Circulation 1986;73:460. Yock PG, Schnittger I, Popp RL. Is continuous wave Doppler too sensitive in diagnosing pathologic valvular regurgitation [Abstract]. Circulation 19&;70 (suppl II)%-381. - Takao S. Mivatake K. Izumi S. Kinoshita N. Sakakibara H. Nimura ?. physiologic pulmonary regurgitation detected by the Doppler technique and its differential diagnosis [Abstract]. J Am Co11 Cardiol 1985;5:499. Kostucki W, Vandenbossche J, Friart A, Englert M. Pulsed Doppler regurgitant flow patterns of normal valves. Am J Cardiol 1986;58:309.
Lens changes cardiovascular diabetes
9. Akasaka T, Yoshikawa J, Yoshida K, Okumachi F, Koizumi K, Shiratori K, Takao S, Shakudo M, Kate H. Age-related v&ular regurgitation: a study by pulsed Doppler echocardiography. Circulation 1987;76:262. 10. Weyman AE. Standard plane positions-standard imaging planes. In: Weyman AE. Cross-sectional echocardiography. Philadelphia: Lea & Febiger, publishers, 1982:98. II. Adhar GC, Abbasi AS, Nanda NC. Doppler echocardiography in the assessment of mitral regurgitation and mitral valve prolapse. In: Nanda NC, ed. Doppler echocardiography. New York: Igaku-Shoin, 1985:197. 12. Feigenbaum H. Acquired valvular heart disease. In: Feigenbaum H. Echocardiography. Philadelphia: Lea & Febiger, Publishers 1986:262.
and the incidence of events among persons
with
Previous analysis of data from the Framingham Heart Study and the Framingham Eye Study showed that the “all-cause” death rate for diabetic persons with lens opacities was more than twice that of diabetic persons without lens opacities. Additional follow-up information was used to investigate whether these lens changes were associated with cardiovascular morbidity and mortality in particular. At the eye examination (1973 to 1975) there were 133 diabetic persons with no cardiovascular disease, of whom 41% had lens changes. Of these 133 persons, 57 had at least one cardiovascular event by the time of the most recent follow-up examination (1981 to 1983). Regression analyses suggested an increased risk of cardiovascular events among diabetic persons with lens changes (incidence rate ratio = 1.8; p = 0.07). Specifically there were associations of lens changes with the development of congestive heart failure (incidence rate ratio = 3.6; p = 0.01) and coronary heart disease (incidence rate ratio = 2.4; p = 0.08). Thus lens changes, in addition to being early prognostic signs of mortality, appear to be predictors of cardiovascular disease in adult-onset diabetes. (AM HEART J 1989; 117:642.)
Marvin Robert
J. Podgor, MS, William B. Kannel, MD, Gary H. Cassel, MD, and D. Sperduto, MD. Bethesda and Baltimore, Md., and Boston, Mass.
A significant association (p = 0.001) of lens changes with decreased survival in persons with diabetes was previously noted in the Framingham cohort.’ In particular the death rate for diabetic persons with lens changes was estimated to be more than twice that of diabetic persons without lens changes. Lens From the Biometry and Epidemiology Program, National Eye Institute, Boston University School of Medicine, and Johns Hopkins Hospital, Wilmer Institute. Received for publication Sept. 23, 1988; accepted Nov. , 1988. Reprint requests: Marvin J. Podgor, Biometry and Epidemiology Program, National Eye Institute, Building 31 Room 6A16, National Institutes of Health,
642
Bethesda,
MD
20892.
changes were associated with cardiovascular deaths, cancer deaths, and perhaps deaths from other causes. Cause-specific mortality data are of limited use, however, when interest is in specified morbidity, in part because the cause of death often incompletely describes an individual’s underlying systemic condition. Specific cardiovascular events have been identified in participants of the Framingham Heart Study29 3; with the use of these data we now report the results of an investigation into whether lens changes are associated with cardiovascular morbidity, to which persons with diabetes are particularly
Volume Number
117 3
Lens changes and cardiovascular
Table I. Seven-year cumulative the Framingham Eye Study
incidence
of any cardiovascular
event among
those with
Sex
No. at risk
52-64 65-74 75-85 52-64 65-74 75-85
30 25 6 33 18 21
Men Women
Data presented
as number
of persons at start of follow-up
Percent 7-year cumulative incidence (SE) 43 37 75 24 35 43 and 7-year
cumulative
vulnerable.4 This investigation is based on approximately 2 additional years of follow-up beyond that reported in our survival study.’ METHODS The study population has been described in detail.’ Briefly, the Framingham Heart Study, which began in 1948, was designed to investigate factors associated with the development of cardiovascular diseases in a representative sample of the adult population of Framingham, Massachusetts.2 The participants have been reexamined approximately every 2 years. From 1973 to 1975, during the twelfth and thirteenth examination cycles of the heart study, the Framingham Eye Study was conducted on survivors of the heart study cohort who volunteered for examination. The principal purposes of the eye study were to estimate prevalences and identify risk factors for major ocular diseases. The study protocol specified standardized observations by trained examiners5 Each participant in the eye study had an examination consisting of a medical history, determination of visual acuity, slit-lamp examination of the anterior segment including the lens, and indirect and direct ophthalmoscopy of the fundus.5*6 A lens change was defined as any of the following: cortical cuneiform opacities, decreased nuclear lucency, posterior subcapsular changes, other late lens changes, or aphakia of senile cause.’ Diabetes was assessed by the following criteria: (1) a casual blood glucose level greater than 160 mg/dl at two or more heart study examinations, (2) a blood glucose level greater than 205 mg/dl 1 hour after a 50 gm oral glucose load at the eleventh heart study examination, (3) a history of a glucose tolerance test result obtained elsewhere indicating diabetes, or (4) a history of the use of oral hypoglycemic agents or insulin prescribed by a physician. More than 95% of the persons with diabetes had adult onset disease (Milton RC, Wilson PWF, Ferris FL. Unpublished data). Cardiovascular events are grouped into four major categories: coronary heart disease (including myocardial
no cardiovascular
643
events at
Persons without diabetes
Persons with diabetes Age at eye examination (Yr)
events in diabetes
No, at risk
(9) (10) (20) (8) (12) (11) incidence
16(2)
369 147 53 496 265 98 (standard
error)
Percent 7-year cumulative incidence (SE) 27 (4) 37 (7) 11 (1)
16 (2) 28 (5) in percent.
infarction, angina pectoris, sudden death, or coronary cerebrovascular accident, intermittent insufficiency), claudication, and congestive heart failure.3 Of the 207 persons with diabetes and the 1738 persons without diabetes in our previous analysis,’ 133 (64%) and 1428 (82%), respectively, had not had any cardiovascular event by the time of the eye study. The 133 persons with diabetes who were at risk of an initial cardiovascular event are the primary subjects of the present analysis. Follow-up information was available through the seventeenth heart study biennial examination, corresponding to years 1981 to 1983. For each person follow-up times were computed for each of the four major types of cardiovascular events. If a person had one or more events of a particular type, the number of days was calculated from the eye study examination until the earliest occurrence of that type of event. If a person did not have an event of a particular type, the last follow-up date associated with that event type was taken to be the date of the seventeenth heart study examination (either the actual date of the examination or, if the person did not attend the seventeenth examination, the scheduled examination date) or the date of death, if death occurred before or within the window of the seventeenth examination. The exact dates of occurrence of some events (unrecognized myocardial infarction, angina pectoris, and intermittent claudication) were not known, the event having been identified at a biennial heart study examination. In such instances event dates were interpolated.a The Cox proportional hazards regression model9 was used to compare incidence rates of cardiovascular events in persons with and without lens changes. Appropriateness of the proportional hazards model was checked by graphic and analytic methods.gs1o The likelihood ratio test was used to test the significance of the regression coeihcients. All tests were two tailed. Minimum follow-up time was 7 years, and maximum time was just over 10 years. RESULTS
Results are presented for the occurrence of “any” cardiovascular event (that is, the first occurrence of
644
Podgor
vi i!!
American
Table II. Character&k of 133 events at, the eye examination
No. of persons ..-..-.--.
Group _ .__. ~_~.._..._..
78 55
No lens changes Lens changes All persona with diabetes *SD,
Standard
133
with
persons
diabetes
in the Framingham
Mean age at eye examination No. of in years women (SD) * (%) ,..-.____-. .__ -~ _---.62.9 (6.3) 70.4 (8.0) 66.0 (7.9)
Eye Study
Mean duration of diabetes at eye examination in years (SD)
---
who had no cardiovascular No. with retinopathy at eye eramination I%)
4.9 (5.1) 7.2 (6.8) 5.9 (6.0)
33 (42) 39 (71) 72 (54)
March 1999 Heart Journal
No. developing cardiovascular events (%I 28 (36) 29 (53) 57 (43)
7 (9) 12 (22) 19 (14)
deviation
Table III. Proportional hazards analysis of incidence of any cardiovascular event among the Framingham Eye Study who had no cardiovascular events at the eye examination 133 Persons Regression Variable
Age(yr) Sex (1 = men, 2 = women) Lens changes (0 = absent, 1 = present) *t Statistic = regression coefficient/sLandard tp value from likelihood ratio test. ZNinety-five percent confidence interval.
coefficient (SE)
0.0132 -0.6384 0.6100
(0.0200) (0.2875) (0.3321)
with
diabetes
t statistic* 0.66 -2.22 1.84 (p = 0.07)t
133 persons with diabetes in .-__-
.-.
(57 events) Estimated rate
incidence ratio
1.01 0.53 (0.9;?.53)t
error
coronary heart disease, cerebrovascular accident, intermittent claudication, or congestive heart failure) and for specific cardiovascular events. Table I shows the 7-year cumulative incidence of any cardiovascular event by sex, age, and diabetes status. Cumulative incidence is presented at 7 years, the minimum follow-up time. The risk for cardiovascular events tends to increase with age and is higher among men than women. The increased risk for persons with diabetes is readily apparent at all ages in both sexes. Stratification of Table I by lens change status results in small cell sizes. Furthermore, such a tabulation presents data at a single time point (year 7 of follow-up). The Cox proportional hazards model uses the exact times of events through the maximum follow-up time and provides an estimate of the ratio of incidence rates while adjusting for confounding variables. A Cox regression model including age, sex, diabetes status, and lens change status and their interactions was fitted for incidence of any cardiovascular event among the 1561 persons with no cardiovascular events at the eye study. This yielded significant interaction terms for age and diabetes (p = 0.04), age and lens changes (p = 0.03), and diabetes and
lens changes (p = 0.08). We then stratified by age groups used in previous eye study publications5-? 52 to 64, 65 to 74, and 75 to 85 years. Results were similar in the two younger age groups so these groups were pooled. For persons aged 52 to 74 years the only significant interaction was that of diabetes and lens changes (p = 0.04). There were no significant interactions in the older age group. Thus, at least in the younger age groups, there appeared to be a differential association of lens changes and incidence of a cardiovascular event between persons with and without diabetes. No association of lens changes and incidence of cardiovascular events was found among persons without diabetes, but as will be shown there is evidence of an association between lens changes and incidence of cardiovascular events among persons with diabetes. Because there was no significant interaction of age and lens changes among persons with diabetes, we will report results based on all 133 persons with diabetes who initially had no cardiovascular events. Table II presents characteristics of these 133 persons with diabetes according to lens change status. The 55 persons with lens changes tended to be older, more often women, and had a somewhat longer duration of diabetes than the 78 persons
vohllo Numb.1
117 3
without lens changes. In addition, those with lens changes also had a somewhat higher prevalence of retinopathy. Among these 133 persons with diabetes, 57 (43 % ) subsequently had at least one cardiovascular event by the time of the most recent follow-up examination. The observed cumulative incidence curves are shown in Fig. 1. The curves suggest differences in development of cardiovascular events between diabetic persons with and without lens changes. Indeed results of proportional hazards analysis adjusted for differences in age and sex indicated that the incidence rate among diabetic persons with lens changes exceeded that among diabetic persons without lens changes (estimated incidence rate ratio = 1.84; 95% confidence interval = 0.96 to 3.53), with a significance level equal to 0.07 (Table III). Because we did not find duration of diabetes and retinopathy to be significant predictors of occurrence of cardiovascular events, perhaps because of the relatively short duration and presence of mild or no retinopathy in these persons, we have not included these variables in the results. The results were unchanged when duration of diabetes and presence or absence of retinopathy were taken into account (data not presented). As in our previously reported survival analysis,* no other variables measured at the time of the eye study and found to be associated with senile cataracts” were also associated with both lens changes and development of cardiovascular events. For clarity of presentation we have provided only the observed curves in Fig. 1. When the curves are adjusted for age and sex they are somewhat wider apart than the observed curves. This is because of the higher proportion of women in the lens change group (Table II) and the lower incidence of cardiovascular events in women than in men (Table I). Although those with lens changes tended to be older than those without lens changes (Table II), so that age-adjustment would be expected to pull the curves closer together, there is a minimal effect of age beyond that of lens changes in the development of cardiovascular events (Table III). Under the proportional hazards assumption the ratio of the incidence rate for diabetic persons with lens changes to the incidence rate for diabetic persons without lens changes will be constant over time. This implies, among other things, that the curves become progressively wider apart beginning immediately after the start of follow-up. It may be noted, however, that the curves in Fig. 1 are close together until about 4 years of follow-up. Indeed model checking for the proportional hazards assumption suggests (at about the 0.10 significance
Lens changes
;3 ii 0
and cardiovascular
events in diabetes
645
loom--
6
80-
8 2
7060-
ANY CARDIOVASCULAR EVENT
iI I -a-
----_I +
YEARS OF FOLLOW-UP Fig.
1. Observed
cumulative
incidence
of any cardiovas-
cular event for diabetic persons with lens changes (dotted line)
and without
lens changes
(solid
line).
level) that the ratio of the incidence rate for diabetic persons with lens changes to the incidence rate for diabetic persons without lens changes may increase with time. This checking was done with the inclusion of a time-dependent covariate in the regression model.gr lo Other graphic methodslo, I2 also yield results that are consistent with the suggestion of an increasing incidence rate ratio with time. Point estimates suggest that for up to 4 years of follow-up the average estimated incidence rate ratio is 1.24; for 4 years and beyond, the average ratio is 2.93. A test that these two ratios differ was significant at about the 0.10 level, a result similar to that obtained from the proportional hazards model checking. Results for the occurrence of specific cardiovascular events are limited by small numbers (Table IV). However, an association of lens changes with the development of coronary heart disease and congestive heart failure is indicated with substantial rate ratios. No such associations are suggested for cerebrovascular accident and intermittent claudication, although the confidence intervals are wide. Figs. 2 and 3 present the observed cumulative incidences for coronary heart disease and congestive heart failure, respectively. As for cardiovascular events in general (Fig. l), the curves adjusted for age and sex are wider apart than the depicted observed curves of Figs. 2 and 3. However, unlike that for total cardiovascular events, we found no evidence of changing incidence rate ratios over time for specific cardiovascular events. When an event of interest was defined to be the occurrence of either coronary heart disease or congestive heart failure, the estimated incidence
646
Podgor
et a!.
Table IV. Proportional hazards analysis diabetes in the Framingham Eye Study
of incidence of specific cardiovascular events among 133 persons with who had no cardiovascular events at the eye examination No. of euents among 78 persons without lens changes
Total events
Event
25 19 11 20
Coronary heart disease Cerebrovascular accident Intermittent claudication Congestive heart failure CI = confidence interval. ‘Incidence rate ratio estimates
fbrch 1969 Heart Journal
Amerkan
change
No. of events among 55 persons with lens changes
13 8 6 7
ill incidence
rate
of the specified
Estimated incidence rate ratio (95% CZ)*
12 11 5 13
event
associated
with
lens
2.39 1.23 1.37 3.65
changes,
adjusted
(0.92, (0.42, (0.33, (1.27,
p
6.22) 3.64) 5.64) 10.49)
for age and
value 0.08 0.70 0.66 0.01
sex.
Table V. Proportional hazards analysis of incidence of any cardiovascular event among 133 persons with diabetes the Framingham Eye Study who had no cardiovascular events at the eye examination according to type of lens change Type
of lens
change*
Cortical Nuclear Posterior
subcapsular
No.
of persons 35 32 13
*With or without other types of lens change. tIncidence rate ratio estimates change in incidence Reference group for each type of lens change contains
No. of events
Estimated rate
22 15 8
rate of any cardiovascular event associated with the 78 persons (28 events) without lens changes.
rate ratio was 2.74 (95% confidence interval = 1.20 to 6.27), with a significance level of 0.02. The results for proportional hazards analyses of any cardiovascular event by type of lens change are presented in Table V. For each analysis the incidence of cardiovascular events for the specified lens change was compared with that of the 78 persons without any lens changes. Results are presented only for cortical, nuclear, and posterior subcapsular opacities, as other lens changes and aphakia occurred in few persons. Although the point estimates of the incidence rate ratios suggest an association of each type of lens change with development of a cardiovascular event, only the ratio for cortical changes was statistically significant (Table V). Persons who had more than one type of lens change are included in two or perhaps all three of the regressions in Table V. Analyses for persons with only those specified lens changes resulted in smaller sample sizes but yielded results similar to those in Table V (data not presented). Diabetes data from participants in the eye study have recently been reviewed.13 The first two criteria in the definition of diabetes (see Methods) have been replaced by the criterion casual serum glucose level greater than 200 mg/dl at two or more heart
incidence ratiot
p value
2.61 1.39 1.79
specified
lens
in
0.01 0.42 0.29
changes,
adjusted
for age and
sex.
study examinations. The last two criteria in the definition were retained. Of the 133 persons with diabetes in the current analysis, 85 were identified as diabetic under the revised definition. There were 41 events among these 85 persons, 16 fewer than among the 133. Proportional hazards regression analyses yielded somewhat reduced incidence rate ratios and lower statistical significance for any and specific cardiovascular events when diabetes was defined by the revised criteria (data not presented). On the other hand, the evidence for an increasing incidence rate ratio with time, for any cardiovascular events, was somewhat stronger (p = 0.04). We have also used blood sugar values from the heart study examination nearest to or preceding the eye study in the entire study group. (More than 91% of these values were taken at the twelfth heart study examination.) We found results similar to those with the use of our original definition of diabetes: there were significant interactions of age and blood sugar, age and lens changes, and blood sugar and lens changes in a Cox regression analysis for any cardiovascular event. With the cutoff values 120 mg/dl or 130 mg/dl, we found no association with lens changes among those with low blood sugar values and suggestions of associations with lens changes
Volume117 Number3
;: g
Lens changes and cardiovascular
100 lOO90 90-
880 8 80-F
647
loo-
CORONARY DISEASE
HEART B 80-
8
70-
g
70-
$ g
6050-
g g
6050-
5-O i= 4 30-
9 5
403020-
5
events in diabetes
20-
r-----------s-r --s-r dJ dJ I I
r----------
I
I
012345678
I
I
2
I 9
I 10
0
YEARS OF FOLLOW-UP
CONGESTIVE CONGESTIVE FAILURE
HEART
--
-r-----
1
2
I-
t--
3 4 5 6 7 YEARS OF FOLLOW-UP
I I I
r----J
8
9
I 10
3. Observed cumulative incidence of congestive heart failure for diabetic personswith lens changes (dotted line) and without lens changes (solid line).
Fig. 2. Observed cumulative incidence of coronary heart diseasefor diabetic persons with lens changes (dotted line) and without lens changes (solid line).
Fig.
among those with “high” blood sugar values so defined (incidence rate ratios were about 1.6). A limitation of this analysis is that the blood sugar values represent one point in time. In particular, some diabetic patients who are receiving treatment may have had low blood sugar values.
event with time, we are limited by our data in explaining this suggestion. One may infer from this finding, however, that lens changes may reflect some insidious cumulative atherogenic aberration of metabolism. Analyses for specific cardiovascular events indicated a strong relationship between lens changes and development of congestive heart failure, a suggested relationship between lens changes and development of coronary heart disease, and a strong relationship between lens changes and development of either of these two events. Perhaps because of small sample sizes, however, no time dependency was detected for specific events. The particularly strong relationship with cardiac failure suggests that lens changes in persons with diabetes may be a marker for diabetic myocardial damage. Of the three types of lens changes investigated, only cortical changes were significantly associated with increased incidence of any cardiovascular event. Although the incidence rate ratios also exceeded one for diabetic persons with nuclear opacities and for diabetic persons with posterior subcapsular opacities (each compared to diabetic persons without any lens changes), these ratios were not statistically significant. Our previous study noted more than a doubling of the death rate for diabetic persons with lens changes compared to the rate for those without lens changes.’ The data from the additional 2 years of follow-up (21 additional deaths, for a total of 87 deaths) strengthened this association between lens changes and mortality. Although we have not presented the updated mortality information in this report, we briefly summarize the results. Specifical-
DISCUSSION
Epidemiologic studies have identified obesity, elevated blood cholesterol levels, systemic hypertension, and cigarette smoking as risk factors for development of cardiovascular disease. Diabetes is also widely recognized as an important risk factor for cardiovascular morbidity and mortality. However, cardiovascular risk factors unique to persons with diabetes have not been elucidated.14*15 Therefore, for epidemiologic and possibly clinical purposes, it is important to identify predictors or markers for increased risk of cardiovascular disease specific to diabetes. Results of the present study suggest that lens changes are markers for increased risk of cardiovascular events in adult-onset diabetes. Among the 133 persons with diabetes who initially had no cardiovascular events, lens changes were associated with subsequent development of such events. In particular the incidence rate for the occurrence of “any” cardiovascular event (that is, the first occurrence of coronary heart disease, cerebrovascular accident, intermittent claudication, or congestive heart failure) was about 80% higher among diabetic persons with lens changes than among those without lens changes. Although there was a suggestion of an increasing incidence ratio of any cardiovascular
American
ly, the updated data yielded an estimated death rate ratio of 2.54 Cp = 0.0002) for 107 diabetic persons with lens changes to 100 diabetic persons without lens changes, adjusted for age and sex; the 95% confidence interval for the death rate ratio ranged from 1.53 to 4.22. Furthermore, among all 207 persons with diabetes examined at the eye study, we did not find t,hat the death rate ratio changed over time for all causes of death. For cardiovascular mortality, however, the death rate ratio did appear to be increasing over time; this is in agreement with our finding of the development of cardiovascular events among those diabetic persons who initially had no events. There was a corresponding decreasing death rate rat.io with time for other causes of death. Finally, death rates for persons with diabetes having each of the three types of lens changes were significantly greater than death rates for those without any lens changes for death from any cause. Our study has a number of strengths: (1) ocular examinations and evaluation of cardiovascular events were conducted according to carefully designed protocols; (2) all persons were followed for at least 7 years and some for up to 10 years; and (3) because of close heart study monitoring, there were few, if any, losses to follow-up. Limitations are the numerically infrequent occurrences of specific cardiovascular events (Table IV) and specific lens changes (that is, posterior subcapsular opacities, Table V). Conclusions therefore should be cautiously drawn from these analyses.16 The biologic significance of our findings is difficult to interpret given the complexity of factors implicated in the development of cardiovascular disease and lens changes. A direct cause-effect relationship between the two is biologically implausible. Lens opacities most likely are markers of systemic biochemical or metabolic insults in diabetes. Systemic insults such as elevated blood sugar or drugs may not only cause changes in the lens but also result in dyslipidemia and thrombogenic tendencies that affect cardiovascular status. The results of this study suggest that among persons with diabetes, lens changes, and perhaps specifically cortical opacities, may be markers for arterial and myocardial damage. These data imply that, in addition to being early prognostic signs of mortality, lens changes are also predictors of cardiovascular sequelae in adultonset diabetes.
March 1999 Heart Journat
REFERENCES
1. Podgor MJ, Cassel GH, Kannel WB. Lens changes and survival in a population-based study. N Engl J Med 1985; 313:1438-44. 2. Gordon T, Kannel WB. Introduction and general background: the Framingham study: an epidemiologic investigation of cardiovascular disease (section 1). Washington, DC: Government Printing Ofice, 1968. 3. Shurtleff D. Some characteristics related to the incidence of cardiovascular disease and death: Framingham study, 18year follow-up. In: Kannel WB, Gordon T, eds. The Framingham study: an epidemiologic investigation of cardiovascular disease (section 30). Washington, DC: Government Printing Office, 1974 (DHEW publication, No. (NIH) 74-599). 4. National Diabetes Data Group. Diabetes in America. Diabetes data compiled 1984. Washington, DC: Government Printing Office, 1985 (NIH publication No. 851468). 5. Kahn HA, Leibowitz HM, Ganley JP, et al. The Framingham Eye Study. I. Outline and major prevalence findings. Am J Epidemiol 1977;106:17-32. 6. Leibowitz HM, Krueger DE, Maunder LR, et al. The Framingham Eye Study monograph: an ophthalmological and epidemiological study of cataract, glaucoma, diabetic retinopathy, macular degeneration, and visual acuity in a general population of 2631 adults, 1973-1975. Surv Ophthalmol 1980;24(suppl):355-610. I. Sperduto RD, Seigel D. Senile lens and senile macular changes in a population-based sample. Am J Ophthalmol 1980;90%-91. 8. Sorlie P. Cardiovascular diseases and death following myocardial infarction and angina pectoris: Framingham study, 20-year follow-up. In: Kannel WB, Gordon T, eds. The Framingham study: an epidemiologic investigation of cardiovascular disease (section 32). Washington, DC: Government Printing Office, 1977 (DHEW publication No. (NIH) 771247). 9. Cox DR. Regression models and life-tables (with discussion). J R Stat Sot (B) 1972;34:187-220. 10. Dixon WJ, ed. BMDP statistical software. Berkeley, Calif.: University of California Press, 1983576-94. 11. Kahn HA, Leibowitx HM, Ganley JP, et al. The Framingham Eye Study. II. Association of ophthalmic pathology with single variables previously measured in the Framingham heart study. Am J Epidemiol 1977;106:33-41. 12. Mau J. On a graphical method for the detection of timedependent effects of covariates in survival data. Appl Statist 198635245-55. 13. Hiller R, Sperduto RD, Podgor MJ, et al. Diabetic retinopathy and cardiovascular disease in type II diabetics: the Framingham Heart Study and the Framingham Eye Study. Am J Epidemiol 1988;128:402-9. RJ. The epidemiology of coronary heart disease and 14. Jarrett related factors in the context of diabetes mellitus and impaired glucose tolerance. In: Jarrett RJ, ed. Diabetes and heart disease, Elsevier Science Publishers B V, 1984;1-23. 15. Oglesby P. The primary prevention of coronary heart disease. Cardiol Clin 1985;3:301-6. 16. Johnson ME, Tolley HD, Bryson MC, Goldman AS. Covariate analysis of survival data: a small-sample study of Cox’s model. Biometrics 1982;38:685-98.
Lens changes cardiovascular diabetes
9. Akasaka T, Yoshikawa J, Yoshida K, Okumachi F, Koizumi K, Shiratori K, Takao S, Shakudo M, Kate H. Age-related v&ular regurgitation: a study by pulsed Doppler echocardiography. Circulation 1987;76:262. 10. Weyman AE. Standard plane positions-standard imaging planes. In: Weyman AE. Cross-sectional echocardiography. Philadelphia: Lea & Febiger, publishers, 1982:98. II. Adhar GC, Abbasi AS, Nanda NC. Doppler echocardiography in the assessment of mitral regurgitation and mitral valve prolapse. In: Nanda NC, ed. Doppler echocardiography. New York: Igaku-Shoin, 1985:197. 12. Feigenbaum H. Acquired valvular heart disease. In: Feigenbaum H. Echocardiography. Philadelphia: Lea & Febiger, Publishers 1986:262.
and the incidence of events among persons
with
Previous analysis of data from the Framingham Heart Study and the Framingham Eye Study showed that the “all-cause” death rate for diabetic persons with lens opacities was more than twice that of diabetic persons without lens opacities. Additional follow-up information was used to investigate whether these lens changes were associated with cardiovascular morbidity and mortality in particular. At the eye examination (1973 to 1975) there were 133 diabetic persons with no cardiovascular disease, of whom 41% had lens changes. Of these 133 persons, 57 had at least one cardiovascular event by the time of the most recent follow-up examination (1981 to 1983). Regression analyses suggested an increased risk of cardiovascular events among diabetic persons with lens changes (incidence rate ratio = 1.8; p = 0.07). Specifically there were associations of lens changes with the development of congestive heart failure (incidence rate ratio = 3.6; p = 0.01) and coronary heart disease (incidence rate ratio = 2.4; p = 0.08). Thus lens changes, in addition to being early prognostic signs of mortality, appear to be predictors of cardiovascular disease in adult-onset diabetes. (AM HEART J 1989; 117:642.)
Marvin Robert
J. Podgor, MS, William B. Kannel, MD, Gary H. Cassel, MD, and D. Sperduto, MD. Bethesda and Baltimore, Md., and Boston, Mass.
A significant association (p = 0.001) of lens changes with decreased survival in persons with diabetes was previously noted in the Framingham cohort.’ In particular the death rate for diabetic persons with lens changes was estimated to be more than twice that of diabetic persons without lens changes. Lens From the Biometry and Epidemiology Program, National Eye Institute, Boston University School of Medicine, and Johns Hopkins Hospital, Wilmer Institute. Received for publication Sept. 23, 1988; accepted Nov. , 1988. Reprint requests: Marvin J. Podgor, Biometry and Epidemiology Program, National Eye Institute, Building 31 Room 6A16, National Institutes of Health,
642
Bethesda,
MD
20892.
changes were associated with cardiovascular deaths, cancer deaths, and perhaps deaths from other causes. Cause-specific mortality data are of limited use, however, when interest is in specified morbidity, in part because the cause of death often incompletely describes an individual’s underlying systemic condition. Specific cardiovascular events have been identified in participants of the Framingham Heart Study29 3; with the use of these data we now report the results of an investigation into whether lens changes are associated with cardiovascular morbidity, to which persons with diabetes are particularly
Volume Number
117 3
Lens changes and cardiovascular
Table I. Seven-year cumulative the Framingham Eye Study
incidence
of any cardiovascular
event among
those with
Sex
No. at risk
52-64 65-74 75-85 52-64 65-74 75-85
30 25 6 33 18 21
Men Women
Data presented
as number
of persons at start of follow-up
Percent 7-year cumulative incidence (SE) 43 37 75 24 35 43 and 7-year
cumulative
vulnerable.4 This investigation is based on approximately 2 additional years of follow-up beyond that reported in our survival study.’ METHODS The study population has been described in detail.’ Briefly, the Framingham Heart Study, which began in 1948, was designed to investigate factors associated with the development of cardiovascular diseases in a representative sample of the adult population of Framingham, Massachusetts.2 The participants have been reexamined approximately every 2 years. From 1973 to 1975, during the twelfth and thirteenth examination cycles of the heart study, the Framingham Eye Study was conducted on survivors of the heart study cohort who volunteered for examination. The principal purposes of the eye study were to estimate prevalences and identify risk factors for major ocular diseases. The study protocol specified standardized observations by trained examiners5 Each participant in the eye study had an examination consisting of a medical history, determination of visual acuity, slit-lamp examination of the anterior segment including the lens, and indirect and direct ophthalmoscopy of the fundus.5*6 A lens change was defined as any of the following: cortical cuneiform opacities, decreased nuclear lucency, posterior subcapsular changes, other late lens changes, or aphakia of senile cause.’ Diabetes was assessed by the following criteria: (1) a casual blood glucose level greater than 160 mg/dl at two or more heart study examinations, (2) a blood glucose level greater than 205 mg/dl 1 hour after a 50 gm oral glucose load at the eleventh heart study examination, (3) a history of a glucose tolerance test result obtained elsewhere indicating diabetes, or (4) a history of the use of oral hypoglycemic agents or insulin prescribed by a physician. More than 95% of the persons with diabetes had adult onset disease (Milton RC, Wilson PWF, Ferris FL. Unpublished data). Cardiovascular events are grouped into four major categories: coronary heart disease (including myocardial
no cardiovascular
643
events at
Persons without diabetes
Persons with diabetes Age at eye examination (Yr)
events in diabetes
No, at risk
(9) (10) (20) (8) (12) (11) incidence
16(2)
369 147 53 496 265 98 (standard
error)
Percent 7-year cumulative incidence (SE) 27 (4) 37 (7) 11 (1)
16 (2) 28 (5) in percent.
infarction, angina pectoris, sudden death, or coronary cerebrovascular accident, intermittent insufficiency), claudication, and congestive heart failure.3 Of the 207 persons with diabetes and the 1738 persons without diabetes in our previous analysis,’ 133 (64%) and 1428 (82%), respectively, had not had any cardiovascular event by the time of the eye study. The 133 persons with diabetes who were at risk of an initial cardiovascular event are the primary subjects of the present analysis. Follow-up information was available through the seventeenth heart study biennial examination, corresponding to years 1981 to 1983. For each person follow-up times were computed for each of the four major types of cardiovascular events. If a person had one or more events of a particular type, the number of days was calculated from the eye study examination until the earliest occurrence of that type of event. If a person did not have an event of a particular type, the last follow-up date associated with that event type was taken to be the date of the seventeenth heart study examination (either the actual date of the examination or, if the person did not attend the seventeenth examination, the scheduled examination date) or the date of death, if death occurred before or within the window of the seventeenth examination. The exact dates of occurrence of some events (unrecognized myocardial infarction, angina pectoris, and intermittent claudication) were not known, the event having been identified at a biennial heart study examination. In such instances event dates were interpolated.a The Cox proportional hazards regression model9 was used to compare incidence rates of cardiovascular events in persons with and without lens changes. Appropriateness of the proportional hazards model was checked by graphic and analytic methods.gs1o The likelihood ratio test was used to test the significance of the regression coeihcients. All tests were two tailed. Minimum follow-up time was 7 years, and maximum time was just over 10 years. RESULTS
Results are presented for the occurrence of “any” cardiovascular event (that is, the first occurrence of
644
Podgor
vi i!!
American
Table II. Character&k of 133 events at, the eye examination
No. of persons ..-..-.--.
Group _ .__. ~_~.._..._..
78 55
No lens changes Lens changes All persona with diabetes *SD,
Standard
133
with
persons
diabetes
in the Framingham
Mean age at eye examination No. of in years women (SD) * (%) ,..-.____-. .__ -~ _---.62.9 (6.3) 70.4 (8.0) 66.0 (7.9)
Eye Study
Mean duration of diabetes at eye examination in years (SD)
---
who had no cardiovascular No. with retinopathy at eye eramination I%)
4.9 (5.1) 7.2 (6.8) 5.9 (6.0)
33 (42) 39 (71) 72 (54)
March 1999 Heart Journal
No. developing cardiovascular events (%I 28 (36) 29 (53) 57 (43)
7 (9) 12 (22) 19 (14)
deviation
Table III. Proportional hazards analysis of incidence of any cardiovascular event among the Framingham Eye Study who had no cardiovascular events at the eye examination 133 Persons Regression Variable
Age(yr) Sex (1 = men, 2 = women) Lens changes (0 = absent, 1 = present) *t Statistic = regression coefficient/sLandard tp value from likelihood ratio test. ZNinety-five percent confidence interval.
coefficient (SE)
0.0132 -0.6384 0.6100
(0.0200) (0.2875) (0.3321)
with
diabetes
t statistic* 0.66 -2.22 1.84 (p = 0.07)t
133 persons with diabetes in .-__-
.-.
(57 events) Estimated rate
incidence ratio
1.01 0.53 (0.9;?.53)t
error
coronary heart disease, cerebrovascular accident, intermittent claudication, or congestive heart failure) and for specific cardiovascular events. Table I shows the 7-year cumulative incidence of any cardiovascular event by sex, age, and diabetes status. Cumulative incidence is presented at 7 years, the minimum follow-up time. The risk for cardiovascular events tends to increase with age and is higher among men than women. The increased risk for persons with diabetes is readily apparent at all ages in both sexes. Stratification of Table I by lens change status results in small cell sizes. Furthermore, such a tabulation presents data at a single time point (year 7 of follow-up). The Cox proportional hazards model uses the exact times of events through the maximum follow-up time and provides an estimate of the ratio of incidence rates while adjusting for confounding variables. A Cox regression model including age, sex, diabetes status, and lens change status and their interactions was fitted for incidence of any cardiovascular event among the 1561 persons with no cardiovascular events at the eye study. This yielded significant interaction terms for age and diabetes (p = 0.04), age and lens changes (p = 0.03), and diabetes and
lens changes (p = 0.08). We then stratified by age groups used in previous eye study publications5-? 52 to 64, 65 to 74, and 75 to 85 years. Results were similar in the two younger age groups so these groups were pooled. For persons aged 52 to 74 years the only significant interaction was that of diabetes and lens changes (p = 0.04). There were no significant interactions in the older age group. Thus, at least in the younger age groups, there appeared to be a differential association of lens changes and incidence of a cardiovascular event between persons with and without diabetes. No association of lens changes and incidence of cardiovascular events was found among persons without diabetes, but as will be shown there is evidence of an association between lens changes and incidence of cardiovascular events among persons with diabetes. Because there was no significant interaction of age and lens changes among persons with diabetes, we will report results based on all 133 persons with diabetes who initially had no cardiovascular events. Table II presents characteristics of these 133 persons with diabetes according to lens change status. The 55 persons with lens changes tended to be older, more often women, and had a somewhat longer duration of diabetes than the 78 persons
vohllo Numb.1
117 3
without lens changes. In addition, those with lens changes also had a somewhat higher prevalence of retinopathy. Among these 133 persons with diabetes, 57 (43 % ) subsequently had at least one cardiovascular event by the time of the most recent follow-up examination. The observed cumulative incidence curves are shown in Fig. 1. The curves suggest differences in development of cardiovascular events between diabetic persons with and without lens changes. Indeed results of proportional hazards analysis adjusted for differences in age and sex indicated that the incidence rate among diabetic persons with lens changes exceeded that among diabetic persons without lens changes (estimated incidence rate ratio = 1.84; 95% confidence interval = 0.96 to 3.53), with a significance level equal to 0.07 (Table III). Because we did not find duration of diabetes and retinopathy to be significant predictors of occurrence of cardiovascular events, perhaps because of the relatively short duration and presence of mild or no retinopathy in these persons, we have not included these variables in the results. The results were unchanged when duration of diabetes and presence or absence of retinopathy were taken into account (data not presented). As in our previously reported survival analysis,* no other variables measured at the time of the eye study and found to be associated with senile cataracts” were also associated with both lens changes and development of cardiovascular events. For clarity of presentation we have provided only the observed curves in Fig. 1. When the curves are adjusted for age and sex they are somewhat wider apart than the observed curves. This is because of the higher proportion of women in the lens change group (Table II) and the lower incidence of cardiovascular events in women than in men (Table I). Although those with lens changes tended to be older than those without lens changes (Table II), so that age-adjustment would be expected to pull the curves closer together, there is a minimal effect of age beyond that of lens changes in the development of cardiovascular events (Table III). Under the proportional hazards assumption the ratio of the incidence rate for diabetic persons with lens changes to the incidence rate for diabetic persons without lens changes will be constant over time. This implies, among other things, that the curves become progressively wider apart beginning immediately after the start of follow-up. It may be noted, however, that the curves in Fig. 1 are close together until about 4 years of follow-up. Indeed model checking for the proportional hazards assumption suggests (at about the 0.10 significance
Lens changes
;3 ii 0
and cardiovascular
events in diabetes
645
loom--
6
80-
8 2
7060-
ANY CARDIOVASCULAR EVENT
iI I -a-
----_I +
YEARS OF FOLLOW-UP Fig.
1. Observed
cumulative
incidence
of any cardiovas-
cular event for diabetic persons with lens changes (dotted line)
and without
lens changes
(solid
line).
level) that the ratio of the incidence rate for diabetic persons with lens changes to the incidence rate for diabetic persons without lens changes may increase with time. This checking was done with the inclusion of a time-dependent covariate in the regression model.gr lo Other graphic methodslo, I2 also yield results that are consistent with the suggestion of an increasing incidence rate ratio with time. Point estimates suggest that for up to 4 years of follow-up the average estimated incidence rate ratio is 1.24; for 4 years and beyond, the average ratio is 2.93. A test that these two ratios differ was significant at about the 0.10 level, a result similar to that obtained from the proportional hazards model checking. Results for the occurrence of specific cardiovascular events are limited by small numbers (Table IV). However, an association of lens changes with the development of coronary heart disease and congestive heart failure is indicated with substantial rate ratios. No such associations are suggested for cerebrovascular accident and intermittent claudication, although the confidence intervals are wide. Figs. 2 and 3 present the observed cumulative incidences for coronary heart disease and congestive heart failure, respectively. As for cardiovascular events in general (Fig. l), the curves adjusted for age and sex are wider apart than the depicted observed curves of Figs. 2 and 3. However, unlike that for total cardiovascular events, we found no evidence of changing incidence rate ratios over time for specific cardiovascular events. When an event of interest was defined to be the occurrence of either coronary heart disease or congestive heart failure, the estimated incidence
646
Podgor
et a!.
Table IV. Proportional hazards analysis diabetes in the Framingham Eye Study
of incidence of specific cardiovascular events among 133 persons with who had no cardiovascular events at the eye examination No. of euents among 78 persons without lens changes
Total events
Event
25 19 11 20
Coronary heart disease Cerebrovascular accident Intermittent claudication Congestive heart failure CI = confidence interval. ‘Incidence rate ratio estimates
fbrch 1969 Heart Journal
Amerkan
change
No. of events among 55 persons with lens changes
13 8 6 7
ill incidence
rate
of the specified
Estimated incidence rate ratio (95% CZ)*
12 11 5 13
event
associated
with
lens
2.39 1.23 1.37 3.65
changes,
adjusted
(0.92, (0.42, (0.33, (1.27,
p
6.22) 3.64) 5.64) 10.49)
for age and
value 0.08 0.70 0.66 0.01
sex.
Table V. Proportional hazards analysis of incidence of any cardiovascular event among 133 persons with diabetes the Framingham Eye Study who had no cardiovascular events at the eye examination according to type of lens change Type
of lens
change*
Cortical Nuclear Posterior
subcapsular
No.
of persons 35 32 13
*With or without other types of lens change. tIncidence rate ratio estimates change in incidence Reference group for each type of lens change contains
No. of events
Estimated rate
22 15 8
rate of any cardiovascular event associated with the 78 persons (28 events) without lens changes.
rate ratio was 2.74 (95% confidence interval = 1.20 to 6.27), with a significance level of 0.02. The results for proportional hazards analyses of any cardiovascular event by type of lens change are presented in Table V. For each analysis the incidence of cardiovascular events for the specified lens change was compared with that of the 78 persons without any lens changes. Results are presented only for cortical, nuclear, and posterior subcapsular opacities, as other lens changes and aphakia occurred in few persons. Although the point estimates of the incidence rate ratios suggest an association of each type of lens change with development of a cardiovascular event, only the ratio for cortical changes was statistically significant (Table V). Persons who had more than one type of lens change are included in two or perhaps all three of the regressions in Table V. Analyses for persons with only those specified lens changes resulted in smaller sample sizes but yielded results similar to those in Table V (data not presented). Diabetes data from participants in the eye study have recently been reviewed.13 The first two criteria in the definition of diabetes (see Methods) have been replaced by the criterion casual serum glucose level greater than 200 mg/dl at two or more heart
incidence ratiot
p value
2.61 1.39 1.79
specified
lens
in
0.01 0.42 0.29
changes,
adjusted
for age and
sex.
study examinations. The last two criteria in the definition were retained. Of the 133 persons with diabetes in the current analysis, 85 were identified as diabetic under the revised definition. There were 41 events among these 85 persons, 16 fewer than among the 133. Proportional hazards regression analyses yielded somewhat reduced incidence rate ratios and lower statistical significance for any and specific cardiovascular events when diabetes was defined by the revised criteria (data not presented). On the other hand, the evidence for an increasing incidence rate ratio with time, for any cardiovascular events, was somewhat stronger (p = 0.04). We have also used blood sugar values from the heart study examination nearest to or preceding the eye study in the entire study group. (More than 91% of these values were taken at the twelfth heart study examination.) We found results similar to those with the use of our original definition of diabetes: there were significant interactions of age and blood sugar, age and lens changes, and blood sugar and lens changes in a Cox regression analysis for any cardiovascular event. With the cutoff values 120 mg/dl or 130 mg/dl, we found no association with lens changes among those with low blood sugar values and suggestions of associations with lens changes
Volume117 Number3
;: g
Lens changes and cardiovascular
100 lOO90 90-
880 8 80-F
647
loo-
CORONARY DISEASE
HEART B 80-
8
70-
g
70-
$ g
6050-
g g
6050-
5-O i= 4 30-
9 5
403020-
5
events in diabetes
20-
r-----------s-r --s-r dJ dJ I I
r----------
I
I
012345678
I
I
2
I 9
I 10
0
YEARS OF FOLLOW-UP
CONGESTIVE CONGESTIVE FAILURE
HEART
--
-r-----
1
2
I-
t--
3 4 5 6 7 YEARS OF FOLLOW-UP
I I I
r----J
8
9
I 10
3. Observed cumulative incidence of congestive heart failure for diabetic personswith lens changes (dotted line) and without lens changes (solid line).
Fig. 2. Observed cumulative incidence of coronary heart diseasefor diabetic persons with lens changes (dotted line) and without lens changes (solid line).
Fig.
among those with “high” blood sugar values so defined (incidence rate ratios were about 1.6). A limitation of this analysis is that the blood sugar values represent one point in time. In particular, some diabetic patients who are receiving treatment may have had low blood sugar values.
event with time, we are limited by our data in explaining this suggestion. One may infer from this finding, however, that lens changes may reflect some insidious cumulative atherogenic aberration of metabolism. Analyses for specific cardiovascular events indicated a strong relationship between lens changes and development of congestive heart failure, a suggested relationship between lens changes and development of coronary heart disease, and a strong relationship between lens changes and development of either of these two events. Perhaps because of small sample sizes, however, no time dependency was detected for specific events. The particularly strong relationship with cardiac failure suggests that lens changes in persons with diabetes may be a marker for diabetic myocardial damage. Of the three types of lens changes investigated, only cortical changes were significantly associated with increased incidence of any cardiovascular event. Although the incidence rate ratios also exceeded one for diabetic persons with nuclear opacities and for diabetic persons with posterior subcapsular opacities (each compared to diabetic persons without any lens changes), these ratios were not statistically significant. Our previous study noted more than a doubling of the death rate for diabetic persons with lens changes compared to the rate for those without lens changes.’ The data from the additional 2 years of follow-up (21 additional deaths, for a total of 87 deaths) strengthened this association between lens changes and mortality. Although we have not presented the updated mortality information in this report, we briefly summarize the results. Specifical-
DISCUSSION
Epidemiologic studies have identified obesity, elevated blood cholesterol levels, systemic hypertension, and cigarette smoking as risk factors for development of cardiovascular disease. Diabetes is also widely recognized as an important risk factor for cardiovascular morbidity and mortality. However, cardiovascular risk factors unique to persons with diabetes have not been elucidated.14*15 Therefore, for epidemiologic and possibly clinical purposes, it is important to identify predictors or markers for increased risk of cardiovascular disease specific to diabetes. Results of the present study suggest that lens changes are markers for increased risk of cardiovascular events in adult-onset diabetes. Among the 133 persons with diabetes who initially had no cardiovascular events, lens changes were associated with subsequent development of such events. In particular the incidence rate for the occurrence of “any” cardiovascular event (that is, the first occurrence of coronary heart disease, cerebrovascular accident, intermittent claudication, or congestive heart failure) was about 80% higher among diabetic persons with lens changes than among those without lens changes. Although there was a suggestion of an increasing incidence ratio of any cardiovascular
American
ly, the updated data yielded an estimated death rate ratio of 2.54 Cp = 0.0002) for 107 diabetic persons with lens changes to 100 diabetic persons without lens changes, adjusted for age and sex; the 95% confidence interval for the death rate ratio ranged from 1.53 to 4.22. Furthermore, among all 207 persons with diabetes examined at the eye study, we did not find t,hat the death rate ratio changed over time for all causes of death. For cardiovascular mortality, however, the death rate ratio did appear to be increasing over time; this is in agreement with our finding of the development of cardiovascular events among those diabetic persons who initially had no events. There was a corresponding decreasing death rate rat.io with time for other causes of death. Finally, death rates for persons with diabetes having each of the three types of lens changes were significantly greater than death rates for those without any lens changes for death from any cause. Our study has a number of strengths: (1) ocular examinations and evaluation of cardiovascular events were conducted according to carefully designed protocols; (2) all persons were followed for at least 7 years and some for up to 10 years; and (3) because of close heart study monitoring, there were few, if any, losses to follow-up. Limitations are the numerically infrequent occurrences of specific cardiovascular events (Table IV) and specific lens changes (that is, posterior subcapsular opacities, Table V). Conclusions therefore should be cautiously drawn from these analyses.16 The biologic significance of our findings is difficult to interpret given the complexity of factors implicated in the development of cardiovascular disease and lens changes. A direct cause-effect relationship between the two is biologically implausible. Lens opacities most likely are markers of systemic biochemical or metabolic insults in diabetes. Systemic insults such as elevated blood sugar or drugs may not only cause changes in the lens but also result in dyslipidemia and thrombogenic tendencies that affect cardiovascular status. The results of this study suggest that among persons with diabetes, lens changes, and perhaps specifically cortical opacities, may be markers for arterial and myocardial damage. These data imply that, in addition to being early prognostic signs of mortality, lens changes are also predictors of cardiovascular sequelae in adultonset diabetes.
March 1999 Heart Journat
REFERENCES
1. Podgor MJ, Cassel GH, Kannel WB. Lens changes and survival in a population-based study. N Engl J Med 1985; 313:1438-44. 2. Gordon T, Kannel WB. Introduction and general background: the Framingham study: an epidemiologic investigation of cardiovascular disease (section 1). Washington, DC: Government Printing Ofice, 1968. 3. Shurtleff D. Some characteristics related to the incidence of cardiovascular disease and death: Framingham study, 18year follow-up. In: Kannel WB, Gordon T, eds. The Framingham study: an epidemiologic investigation of cardiovascular disease (section 30). Washington, DC: Government Printing Office, 1974 (DHEW publication, No. (NIH) 74-599). 4. National Diabetes Data Group. Diabetes in America. Diabetes data compiled 1984. Washington, DC: Government Printing Office, 1985 (NIH publication No. 851468). 5. Kahn HA, Leibowitz HM, Ganley JP, et al. The Framingham Eye Study. I. Outline and major prevalence findings. Am J Epidemiol 1977;106:17-32. 6. Leibowitz HM, Krueger DE, Maunder LR, et al. The Framingham Eye Study monograph: an ophthalmological and epidemiological study of cataract, glaucoma, diabetic retinopathy, macular degeneration, and visual acuity in a general population of 2631 adults, 1973-1975. Surv Ophthalmol 1980;24(suppl):355-610. I. Sperduto RD, Seigel D. Senile lens and senile macular changes in a population-based sample. Am J Ophthalmol 1980;90%-91. 8. Sorlie P. Cardiovascular diseases and death following myocardial infarction and angina pectoris: Framingham study, 20-year follow-up. In: Kannel WB, Gordon T, eds. The Framingham study: an epidemiologic investigation of cardiovascular disease (section 32). Washington, DC: Government Printing Office, 1977 (DHEW publication No. (NIH) 771247). 9. Cox DR. Regression models and life-tables (with discussion). J R Stat Sot (B) 1972;34:187-220. 10. Dixon WJ, ed. BMDP statistical software. Berkeley, Calif.: University of California Press, 1983576-94. 11. Kahn HA, Leibowitx HM, Ganley JP, et al. The Framingham Eye Study. II. Association of ophthalmic pathology with single variables previously measured in the Framingham heart study. Am J Epidemiol 1977;106:33-41. 12. Mau J. On a graphical method for the detection of timedependent effects of covariates in survival data. Appl Statist 198635245-55. 13. Hiller R, Sperduto RD, Podgor MJ, et al. Diabetic retinopathy and cardiovascular disease in type II diabetics: the Framingham Heart Study and the Framingham Eye Study. Am J Epidemiol 1988;128:402-9. RJ. The epidemiology of coronary heart disease and 14. Jarrett related factors in the context of diabetes mellitus and impaired glucose tolerance. In: Jarrett RJ, ed. Diabetes and heart disease, Elsevier Science Publishers B V, 1984;1-23. 15. Oglesby P. The primary prevention of coronary heart disease. Cardiol Clin 1985;3:301-6. 16. Johnson ME, Tolley HD, Bryson MC, Goldman AS. Covariate analysis of survival data: a small-sample study of Cox’s model. Biometrics 1982;38:685-98.