Distribution of intraocular pressure in an elderly Chinese population, Shihpai, Taiwan (N = 1292).
Relationship between intraocular pressure and age with adjustment for confounding factors.
Lin H, Hsu W, Chou P, Liu CJ, Chou JC, Tsai S, Cheng C. Intraocular Pressure Measured With a Noncontact Tonometer in an Elderly Chinese PopulationThe Shihpai Eye Study. Arch Ophthalmol. 2005;123(3):381-386. doi:10.1001/archopht.123.3.381
To determine the distribution of intraocular pressure (IOP) as measured by a noncontact tonometer (NCT) and risk factors responsible for ocular hypertension in elderly Chinese people.
Population-based study of randomly sampled Chinese people 65 years and older in Shihpai, Taipei, Taiwan.
Main Outcome Measures
Participants completed an interview and underwent a physical examination and a standardized ophthalmic examination, including IOP measurement with the NCT. People with a history of glaucoma were excluded. Risk factors were assessed using multivariate regression analysis.
Of 1361 study participants examined, 1292 (95.4%) had no history of glaucoma and were therefore included in the study. Their mean ± SD IOP was 12.9 ± 3.1 mm Hg. The IOP decreased significantly (P<.001) with age, changing from 13.3 ± 3.0 mm Hg in participants aged 65 to 69 years to 11.6 ± 2.8 mm Hg in those 80 years and older. Women had significantly higher IOP than men (P<.001). In the multivariate regression analysis, decreasing age, female sex, increasing systolic blood pressure, a history of diabetes, and alcohol drinking were significantly associated with increasing IOP.
The distribution of IOP in elderly Chinese people is similar to that found in other East Asian people, with a negative age-IOP relationship. The mean IOP values in this elderly Chinese population were lower than in white people but higher than in Japanese people in similarly aged groups. Establishing the epidemiologic characteristics of IOP with the NCT is important for the mass screening of ocular hypertension.
Intraocular pressure (IOP) is the ocular parameter most commonly associated with glaucoma.1- 3 Factors associated with IOP in epidemiologic surveys of the general population include age, sex, race, hypertension, diabetes, obesity, smoking, drinking, coffee consumption, physical activity, central corneal thickness (CCT), iris color, nuclear opacity, and myopia.1- 12 An increase in IOP with age has been found in many Western populations,2,13- 15 whereas IOP has been confirmed to decrease with aging among Japanese people.5,8,9,16
Currently, an IOP of 21 mm Hg has been adopted as the upper limit of the normal value. Using this level as a cutoff for ocular hypertension, Shiose et al17 found an extremely low prevalence of ocular hypertension and a very high prevalence of low-tension glaucoma in Japan. This finding may reflect a racial peculiarity in the age-specific trend of IOP in the Japanese population. The traditional cutoff value may not be suitable for all populations because it does not take into account the paradoxical relationship between IOP and aging in different races.
The noncontact tonometer (NCT) is a useful instrument for population surveys. Because it requires no corneal contact or topical anesthesia, it not only leaves the cornea undisturbed before further examination but also decreases the risk of disease transmission, which is especially important in areas where the control of communicable diseases is an important concern in public health. The identification of Chinese IOP characteristics using the NCT may make the use of this noncontact technique more efficient.
With respect to the higher prevalence of glaucoma among elderly people, our objective was to use the NCT to evaluate the distribution of IOP while investigating if personal, medical, or other lifestyle factors might exert some level of influence in an elderly Chinese sample in Taiwan.
The Shihpai Eye Study is a population-based survey that investigated vision and age-related eye diseases in an elderly population. The study was conducted between July 1999 and December 2000 in Shihpai, Taipei, Taiwan. Details of the study design and data collection methods have been described elsewhere.18,19 In brief, residents 65 years and older in the Shihpai community were identified through the household registration system. According to the official household registration database, 4750 people 65 years and older were living in the Shihpai community in 1999. After excluding vacant households (n = 658), residents who died before they were contacted (n = 48), and inpatients and paralyzed and disabled people (n = 298), 3746 residents were eligible for the study. Of them, 2045 were randomly selected and invited to participate in the study.
The study was conducted according to the tenets of the Declaration of Helsinki of the World Medical Association regarding scientific research on human subjects, and the study protocol was approved by the institutional review board. Written informed consent was obtained from the participants before enrolling them in the study.
Appropriately trained interviewers first contacted participants and elicited data using a standardized questionnaire that covered sociodemographic characteristics, ocular and medical history, lifestyle, and health behavior. The participants were then invited to further undergo an ophthalmic evaluation. The ophthalmic examination was conducted according to a standardized protocol that covered best-corrected visual acuity, autorefraction (RK-8100; Topcon Corporation, Tokyo, Japan), noncontact tonometry (CT-60 NCT; Topcon Corporation), slitlamp biomicroscopy, indirect ophthalmoscopy, color fundus photography, and a screening visual field test with frequency doubling technology (Humphrey Instruments, San Leandro, Calif).
The IOP measurements were performed in both eyes, and the mean value of 3 successive measurements was accepted as the final result. The automatic alignment system without local anesthetic was used. Lens opacity was graded using slitlamp biomicroscopy with the modified Lens Opacity Classification System III.20 Based on this system, nuclear opacity was graded from score 1 to score 6 in increasing severity. The refractive error was measured with autorefraction and expressed as a spherical equivalent.
Body height in centimeters and body weight in kilograms were measured, with participants lightly clothed and shoeless. The scales were calibrated to zero before use. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters. A mercury sphygmomanometer was used to measure blood pressure. Three consecutive blood pressure readings, separated by at least 5-minute intervals, were obtained from both arms, with participants in a seated position. The higher mean value in either arm was used in the analysis. Hypertension was defined as a mean systolic blood pressure (SBP) of 140 mm Hg or higher, a mean diastolic blood pressure (DBP) of 90 mm Hg or higher, or a history of antihypertensive treatment. Oxygen saturation was measured using a pulse oximeter (model 3301; BCI Inc, Waukesha, Wisc), which displays fractional oxygen saturation. Health behavior indicators were based on self-reported information. Individual smoking and alcohol behaviors were described in the records as current, former, or never. Daily water intake volume was divided into 4 categories, ranging from less than 1000 mL in the lowest group to more than 3000 mL in the highest group.
Participants who had a history of glaucoma or had been treated for glaucoma with either medication or surgery were excluded from the analysis. Those who had undergone cataract surgery were also eliminated from the analysis of the association of IOP with refractive error and nuclear opacity. Since there was a strong correlation between the mean IOP of both eyes (r = 0.78, P<.001), only data for the right eye are presented.
Means ± SDs of IOP were calculated and are presented herein. Univariate associations of IOP with other variables were first detected by the t test or analysis of variance. Statistical significance was set at P<.05. Those variables found to be significant in the univariate analysis were retained for the final forward stepwise multiple linear regression model. All analyses were conducted using SPSS statistical software, version 11.0.0 for Windows (SPSS Inc, Chicago, Ill).
Of the 2045 randomly selected participants, 1361 (66.6%) completed the interview and underwent both physical and ophthalmic examinations. Seven participants (0.3%) could not be contacted during 3 visits for the household interview. A total of 677 participants (33.1%) responded to the questionnaire but did not undergo the ocular examination because they refused or had no spare time available to go to the hospital. In general, participants (mean ± SD age, 72.2 ± 5.1 years) were younger than nonparticipants mean ± SD (age, 74.3 ± 6.0 years), more likely to be male, and had a higher level of education (all P<.001).18
The IOP data were available for 1355 (99.6%) of the 1361 participants who were examined. Sixty-three people (4.6%) were excluded from the analyses because they had a history of glaucoma or had been treated for glaucoma with medication, surgery, or both. Therefore, 1292 people (95.4%) were included in the analyses. The mean ± SD age of the 1292 persons was 72.1 ± 5.0 years, and 777 participants (60.1%) were male.
Figure 1 shows the distribution of IOP, which was slightly skewed to the right. The geometric mean of IOP was 12.9 ± 3.1 mm Hg for the entire study population. An IOP greater than 21 mm Hg was recorded in 1.2% of the participants. Table 1 shows the mean value of IOP, SBP, DBP, and BMI by age group and sex. The IOP significantly decreased with age in both sexes, declining from 13.3 ± 3.0 mm Hg in participants aged 65 to 69 years to 11.6 ± 2.8 mm Hg in those 80 years and older. Women (13.3 ± 3.0 mm Hg) had a significantly higher IOP than men (12.7 ± 3.1 mm Hg). The SBP increased with age in women, whereas DBP decreased with age in men. A significant negative association was also found between BMI and age in men and women.
Table 2 lists the distribution of IOP by potential factors other than age and sex that may be associated with IOP. For continuous variables, the median from all participants was used as a cutoff point to provide the corresponding summary IOP values. The 3 categories of BMI were divided according to the classification of obesity by the World Health Organization.21 The SBP, DBP, pulse rate, and self-reported history of diabetes had a significantly positive association with IOP. Both BMI and alcohol consumption were related to IOP in men only. Neither nuclear sclerosis nor refractive status of the eye was associated with IOP.
Table 3 presents the results of the final stepwise multiple regression analysis. The SBP was entered in the first step of the regression equation, which indicated that SBP had the strongest correlation with IOP. Age was the second contributing factor, followed by diabetes history, sex, and alcohol habit. Overall, the model explained only 8.7% of the variation in IOP, of which 3.6% was contributed by SBP. The BMI did not reach statistical significance (P = .40) when added to the final model.
Age-specific IOP values estimated using the multiple linear regression models that controlled for SBP, diabetes, and alcohol consumption are plotted in Figure 2. The highest levels of mean IOP were separately noted for the 65- to 69-year-old age group in men and the 70- to 74-year-old age group in women. A statistically significant (P<.001) trend toward lower IOP values with age was noticed in both sexes.
The NCT is the most commonly used tonometer in eye clinics in Taiwan. Thus, the detailed characteristics of IOP provided by the present study may provide useful information in screening for ocular hypertension and also help in setting a target IOP for the treatment of glaucoma. Despite the wide acceptance of the Goldmann applanation tonometer, it has the possible disadvantages of being difficult to master and of transferring ocular and systemic disease among patients. By contrast, the NCT is more objective and has the advantages of being a noncontact method. Studies indicate that the NCT is reliable within the normal IOP range.22 The CT-60 model of the NCT instrument used in the present study has been shown to correlate well with Goldmann tonometry.23
Table 4 gives the IOPs measured by the NCT in selected populations.6,10,11,24,25 The mean IOP values measured with an NCT among white participants were between 14.3 and 15.8 mm Hg.24,25 For elderly Japanese participants, the mean IOP values measured using an NCT were between 11.4 and 11.7 mm Hg.9,11 Mean IOPs obtained in the present study were lower than in white participants but higher than in Japanese participants. The finding of lower mean IOP in East Asian participants than white participants can be further supported by comparing other epidemiologic reports that used either Schiotz tonometry12 or Goldmann tonometry.1,3,7,26
With regard to ocular hypertension, an IOP of 21 mm Hg is generally used as the cutoff. This value stands for the mean + 2 SDs of IOP in populations, and 97.5% of healthy individuals should be included within mean + 2 SDs on one side. However, if IOP is measured by the NCT, using 21 mm Hg as the cutoff may not be suitable for our elderly Chinese population because it approximates up to the 99th percentile of the IOP distribution. Therefore, it may be more appropriate to reset the cutoff IOP value at 19.1 mm Hg (mean + 2 SDs) for elderly Chinese people when IOP is obtained by the NCT.
In the present study, IOP values decreased with advancing age, even when adjusted for other potential confounders. The age-specific trend in IOP differs between ethnic groups. Some population-based studies in white populations showed that IOP increased slightly with advancing age,13,15 whereas in others1,3,7 age was not significantly related to IOP when adjusting for other potential confounders. The positive age effect on IOP was similar in a black population in the Barbados Eye Study,2 but a larger age-related increase was observed than in the white populations.2,13,15 By contrast, an inverse association was repeatedly confirmed in many cross-sectional Japanese studies.5,8,9,11,16 The influence of age on IOP in our Chinese population was similar to that found in Japan. To the best of our knowledge, our study provides the first population-based data outside Japan to support the negative relationship between age and IOP in Asian individuals.
Blood pressure and obesity2,3,5,9 should be considered when the association between age and IOP is assessed. We found that SBP was the variable that best correlated with IOP, and an increase in SBP was related to a higher IOP. These findings are similar to those found by other researchers.4,7,14,27 In a few studies,3,5 a correlation also existed between IOP and DBP, but we did not obtain such results. The higher SBP may increase the formation of aqueous humor by ultrafiltration and, through this mechanism, increase IOP.9 The greater importance of SBP than DBP suggests that the height of the pressure wave reaching the eye seems to be a more essential determinant of IOP than the perfusion pressure.9 Obesity was not an independent risk factor in increasing IOP in the final multivariate analysis in the present study. Positive associations between obesity and IOP were found in most white or black populations, such as in the Beaver Dam Eye Study3 and Barbados Eye Study.2 This association was also found in the Japanese population.5,9,11,16 In our study, the BMI effect on IOP was nullified when SBP was taken into account. Therefore, SBP may confound the relationship between BMI and IOP.
The contrast in the age-IOP relationship between Japanese and Western populations may be explained by a lower incidence of hypertension and obesity with advancing age in the Japanese population.9 In our study, the overall prevalence of hypertension and obesity (BMI ≥30.0) was 62.7% and 8.1%, respectively. These data are similar to other reports of elderly people in Taiwan.28,29 The 1999-2000 National Health and Nutrition Examination Survey30,31 reported the prevalence of hypertension as 65% in both sexes and obesity as 31.8% and 35% in men and women, respectively, in persons 60 years and older. The Chinese population in the present study, similar to the Japanese population, contains a lower proportion of obese and hypertensive participants than Western populations. Since the level of IOP is assumed to be maintained by the counterbalance between the IOP-raising factors of hypertension and obesity and the IOP-lowering factor of age,9 the lower proportion of obese and hypertensive participants could probably result in the inverse age-IOP relationship in our study.
Variation in CCT, which was not measured in the present study, should also be considered when examining the relationship between age and IOP.4,8 It has been suggested that CCT has an important influence on IOP measurement, and a positive correlation has been identified between them.4,8,25 Foster et al4 found that in the Chinese Singaporean population, IOP increased significantly with age in univariate analysis. However, when they controlled for CCT, such an association was no longer encountered. Because a decrease in CCT with age has been documented in people of East Asian origin,4,26 it may offer a possible explanation for the observed decrease in IOP with age in Asian individuals. In our study, cohort effects may also have contributed to the decrement in IOP with age; for example, changing dietary habits toward a more Western diet might raise cholesterol levels in younger individuals. A positive correlation between IOP and cholesterol has previously been reported.3
One concern of our study is the relative low response rate of 66.6%, which is similar to that of the Salisbury Eye Evaluation Project.32 Because elderly people generally have a higher frequency of morbidity and disability that hinder travel and motivation to participate, population-based studies in elderly populations are usually limited by a lower response rate. In the present study, although the prevalence of hypertension and diabetes was similar between participants and nonparticipants,18 participants were younger than nonparticipants. Underrepresentation within the older age group probably biases the true age effect on IOP. More important, because higher IOP is a marker of decreased life expectancy,33 mortality should also be considered in such an elderly population. It is possible that participants with higher IOPs are somehow removed from this study owing to earlier death; therefore, our study might have included only elderly people with low IOP.
The association between IOP and lifestyle factors, such as smoking, alcohol consumption, and daily water drinking, is still questionable. In the present study, a positive association between alcohol intake and IOP was found after adjusting for possible confounders in the multivariate model. However, neither smoking nor water drinking was significantly associated with IOP.
In contrast to other studies,1,3,34 little evidence was found in our study for an association between refractive error and IOP. However, we found that IOP values tend to be higher among people with hyperopia than those with emmetropia. Hyperopic eyes are usually associated with smaller eyeballs and thus more crowded anterior segments, which may impede aqueous outflow and produce higher IOP among people with hyperopia. This may also account in part for the relatively high prevalence of angle-closure glaucoma in Chinese populations.35,36
In conclusion, this study provides data on IOP measured with the NCT in an elderly Chinese population with no known history of glaucoma. The age-related decrease in IOP was similar to that reported in other East Asian populations. Increased SBP and diabetes history were associated with increased risk of ocular hypertension. The lower mean IOP value and the negative age-IOP relationship suggest that an IOP level lower than the current standard should be used when defining the disease in epidemiologic studies or setting a target IOP for glaucoma treatment in clinics for elderly Chinese populations.
Correspondence: Ching-Yu Cheng, MD, MPH, Department of Ophthalmology, Taipei Veterans General Hospital, 201 Section 2, Shihpai Road, Taipei 112, Taiwan (firstname.lastname@example.org).
Submitted for Publication: August 6, 2003; final revision received August 9, 2004; accepted August 9, 2004.
Financial Disclosure: None.
Funding/Support: The study was supported by grants from Taipei Veterans General Hospital (VGH 89-404 and VGH 92-136) and by Yen Tjing Ling Medical Foundation (Dr Cheng), Taipei.