Refractive Error, Ocular Biometry, and Lens Opalescence in an Adult Population: The Los Angeles Latino Eye Study

Abstract

purpose. To characterize age- and gender-related differences in refractive error, ocular biometry, and lens opalescence (NOP) in a population-based sample of adult Latinos. Also assessed were the determinants of age-related refractive differences.

methods. Participants in the Los Angeles Latino Eye Study (LALES), a population-based study of Latinos aged 40 years and more, underwent an ophthalmic examination, including ultrasonic measurements of axial length (AL), vitreous chamber depth (VCD), anterior chamber depth (ACD), lens thickness (LT), and noncycloplegic automated and subjective refraction. Corneal curvature/power (CP) was measured using an autorefractor. NOP was graded at the slit lamp by an ophthalmologist using the Lens Opacity Classification System II. Age- and gender-related differences were calculated. Multiple regression models were used to identify the determinants of age-related refractive differences.

results. Of the 6357 LALES participants, 5588 phakic individuals with biometric data were included in this analysis. Older individuals had shallower ACDs, thicker lenses, more NOP, and more hyperopia compared to younger individuals (P < 0.001). There was no age-related difference in AL (P ≥ 0.05). Women had significantly shorter AL, shallower ACD and VCD, than did men (P ≤ 0.01). The strongest determinants of refractive error were AL (primarily VCD) and CP. NOP was a small but significant determinant of refractive error in older individuals.

conclusions. Age- and gender-related differences in ocular biometric, refractive error, and NOP measurements are present in adult Latinos. While the relative contribution of NOP in determining refractive error is small, it is greater in older persons compared to younger individuals.

Visual impairment is becoming an increasingly important public health issue as the world’s population ages. ^{1 2} A leading cause of the increased prevalence of visual impairment with aging is uncorrected refractive error. ³ Refractive error is related to anatomic and functional differences in the eye, particularly in axial length, lens thickness, and lens opalescence. ^{4 5 6 7 8 9 10 11} A detailed biometric evaluation of the eye in an unbiased sample of individuals would help elucidate the underlying reasons for age-related changes in refractive error and provide insight into how to decrease the negative impact of refractive error-related visual impairment on quality of life and activities of daily living. Therefore, population-based data of ocular biometry measurements would be useful in this regard. 

To our knowledge, few population-based studies have provided a comprehensive assessment of ocular biometry and other ocular measurements to help explain refractive changes in the aging eye, ^{10 11} and none have been undertaken in non-Asian populations worldwide. Furthermore, no population-based biometry data exist on adult populations in the United States. Data from East Asia—Mongolia and Singapore—provide conflicting reasons for the age-related shift from myopia toward hyperopia that takes place from the fifth through ninth decades in the average population. ^{10 11} In Singaporean Chinese the age-related hyperopic shift was attributed to changes in axial length and vitreous chamber depth in the 40- to 59-year age group and to lens thickness in those aged 60 years and more. ¹⁰ In contrast, in Mongolians aged 40 years and more, axial length and vitreous chamber depth were the primary determinants of refractive error. ¹¹ In addition, although axial length was smaller in older compared with younger Singaporean Chinese, this pattern was not observed in Mongolians. ^{10 11} Thus, this area of research would benefit from further studies in different populations. 

The Los Angeles Latino Eye Study (LALES), conducted from 2000 to 2003, has obtained data on refractive error, ocular biometry, and lens opalescence in a population-based sample of adult Latinos from La Puente, California, a suburb of Los Angeles. ¹² The purpose of this analysis is to clarify the age- and gender-related differences and the relationships between refraction, ocular biometry, and lens opalescence in a population-based sample of adult Latinos. In addition, this analysis should help further clarify the determinants of age-related changes in adults. 

Methods

Study Cohort

The Los Angeles Latino Eye Study (LALES) is a population-based, cross-sectional study of adult Latinos aged 40 years and older. The survey took place in the City of La Puente in Los Angeles County between 1999 and 2003. This city was chosen because it has a large, stable Latino population, and its socioeconomic demographic profile is similar to that of all Latinos in the US. ¹²  

Interviews and Examination

Details of the study design, sampling plan, and baseline data are reported elsewhere. ¹² Briefly, participants of the study cohort were self-identified Latinos aged 40 years and older living in six census tracts of the City of La Puente, California. After informed consent was obtained, a detailed in-home interview was conducted to determine demographic factors, level of acculturation, ocular and medical histories, various risk factors, and access to medical and ocular care. The Cuellar nine-item acculturation scale, which is based on spoken and written language, was used to measure acculturation (1, lowest acculturation; 5, highest acculturation). All eligible Latinos were invited to attend a local eye examination center (LEEC) for a comprehensive ocular examination to be performed by trained ophthalmologists and technicians, using standardized protocols. Those participants who did not complete a clinical examination at the LEEC were invited to undergo a clinical examination at home to be conducted by a trained ophthalmologist and a trained technician. These participants were not included in the present analyses, since no biometry was performed during these in-home examinations. The Institutional Review Board at the University of Southern California approved the study protocol. The study adhered to the tenets of the Declaration of Helsinki. 

Biometric and Lens Opalescence Measurements

Using an A-scan ultrasound (A-Scan pachymeter, Ultrasonic, Exton, PA), we obtained three measurements for each of four variables: axial length, anterior chamber depth, central corneal thickness, and lens thickness. Vitreous chamber depth was calculated by subtracting the central corneal thickness, anterior chamber depth and lens thickness from the axial length. Lens nuclear opalescence was graded at the slit lamp. The Lens Opacities Classification System II (LOCS II) ¹³ was used to categorize opacities into five nuclear grades (N0, NI, NII, NIII, NIV) of increasing severity, according to photographic standards. 

Refraction and Corneal Power Assessment

Visual acuity for each LALES participant was measured for each eye at 4 m, using standard ETDRS protocols with a modified ETDRS distance chart transilluminated with a chart illuminator (Precision Vision, La Salle, IL). ¹⁴ Those participants who read 55 letters or better unaided were considered emmetropic, and the refractive error was recorded as zero. If the participant read fewer than 55 letters at 4 m in either eye, an automated noncycloplegic refraction and corneal power measurement was performed (Humphrey Autorefractor Model 599; Carl Zeiss Meditec, Dublin, CA). The refraction was further refined by subjective refraction using standard protocols. It is likely that a noncycloplegic refraction could have underestimated hyperopia and underestimated myopia in younger persons. Spherical refractive error was measured to the closest 0.25 D. Cylinder power was measured to the closest 0.25 D and recorded in the negative form. The spherical equivalent (SE; sphere plus half cylinder) was used in all analyses as the measure of refractive error. Three corneal power measurements in the two axes were recorded in terms of dioptric power. The measurements from the two axes were averaged, and the mean of the three measurements was used in these analyses. 

Data Analysis

Demographic, clinical, and ophthalmic data were entered into a database system (Access-98; Microsoft Corp., Redmond, WA) with internal automated range and quality control checks. The biometric data for each eye were analyzed separately. However, because the results for the right and left eyes were similar, we report data from the right eye only in this analysis. Only those participants with complete data variables were included in the analyses. Participants who had undergone right cataract extraction were excluded. The ocular variables in this study were first analyzed by calculating the mean, range, and SD. Mean spherical equivalents, biometric parameters, and LOCS II grade for lens opalescence were calculated in 10-year categories. Age-adjusted statistical differences between genders and trend tests were calculated using the logistic regression analysis, with age as a continuous variable. All confidence intervals presented are 95%. An analysis of variance was conducted separately for males and females to evaluate the variation of different biometric components and refraction across age groups. A trend test was used to assess any significant trends across the 10-year age groups for each variable. To examine possible threshold and nonlinear relationships between different biometric and clinical variables, age, and gender, an iterative, locally weighted, least squares method was used to generate lines of best fit (Lowess fit line). Pearson correlation coefficients were calculated to present the interrelationships between various biometric, refractive, and clinical parameters. We constructed multiple linear regression models with stepwise procedures to assess the contributory effect of each biometric and clinical parameter on refractive error. We assessed the overall contribution of axial length, its components (anterior and vitreous chamber depths and lens thickness) and clinical variables (corneal power and lens opalescence; independent variables) to refractive error (dependent variable) (1) in all participants, after adjusting for age and gender, and (2) for each decade of age from 40 to 80+ years, after adjusting for gender. Standardized regression coefficients (SRCs) and partial correlation coefficients (PCCs) were used to characterize the relative contributory effect of each independent variable on noncycloplegic refractive error. The SRC is calculated by multiplying the original estimate of the regression coefficient with the SD of the independent variable and dividing by the SD of the dependent variable. The SRC is an indication of the relative importance of various independent variables (biometric and clinical variables) with regard to the value of the dependent variable (noncycloplegic refractive error). An SRC with a high absolute value is indicative of its associated independent variable having a high degree of influence on the dependent variable. The PCC is the correlation between the dependent and independent variables, when all other variables are held constant. The PCC expresses the proportion of the total variability in the dependent variable attributable to the independent variable. A PCC with a high value is indicative of the independent variable explaining a high degree of variability in the dependent variable. All analyses were performed on computer (Statistical Analysis System, ver. 8; The SAS Institute, Cary, NC). The graphs for the relationship of biometric, refractive, and clinical variables with age were also created on computer (SPSS statistical software, ver. 11.5; SPSS Inc., Chicago, IL). 

Results

Of the 7789 eligible Latinos, 6870 (88%) completed an in-home interview. Of those who completed an interview, 6357 (82% of those eligible) underwent a clinical examination. Those who completed an in-home interview were, on average, 2 years younger (54.9 ± 11 years) and more likely to be female (58%) than were those who did not complete the in-home interview (56.8 ± 11 years, female 47%, P < 0.0001). 

Biometric data, refractive error, and lens opalescence data were available for 5588 (88%) of those who participated in the clinical examination. Compared with those who were excluded from the analysis, the study subjects were younger (54.2 vs. 61.8 years), less likely to be born outside the United States, less likely to be unemployed (31% vs. 50%), and less likely to have a low level of acculturation (61% vs. 100%; all comparisons P < 0.001). 

Refractive Error

Tables 1 and 2present the mean, SD, and range of noncycloplegic refractive error in Latinos, stratified by gender and age group. Overall, the average refractive error was hyperopic (+0.11D) with a range of −19.0 to +10.6 D in females (mean, +0.18 D) and a range of −14 D to +10 D in men (mean, +0.02 D). In females, the mean refractive error ranged from −0.32 D in those aged 40 to 49 years, to +1.02D in those aged 70 to 79 years and +0.74 D in those aged 80 years and older. In males, the mean refractive error ranged from −0.3 D in those aged 40 to 49 years (similar to the females), to +0.6 D in those aged 70 to 79 years old (on average approximately 0.5D less hyperopic than the women) and −0.3 D in those aged 80 years and older (on average approximately 1 D myopic compared to the women) (Table 2) . The overall difference and the difference at each age group (except the 40 to 49 year age group) between the men and women were significant (P < 0.001). These gender-related differences in refractive error were not statistically significant after adjusting for height (P = 0.79). 

The distribution of mean refractive error for each year is presented in Figure 1 . The data for this (and each subsequent) parameter are plotted by means for each year from 40 to 80+, since a plot that included data for each individual (n = 5588) would add little to illustrate the relationship of the parameter to age. The locally weighted regression lines that best fit the data are also shown in Figure 1 . Overall, for all participants in the 40 to 80+ year age group, each additional year of age correlated with a 0.04-D increase in the spherical equivalent of hyperopia (Table 3) . This hyperopic trend was noted in the 40- to 69-year age group (hyperopic shifts of 0.048 D/yr in the 40–49 year, 0.07 D/yr in the 50–59 year, 0.02 D/yr in the 60–69 year, and 0.001D/yr in the 70–79 year age groups); but a myopic trend was noted in those 80 years and older (−0.07D/yr; Table 3 ). When the data was stratified by gender, there was a significant hyperopic shift in the refractive error in females from age 40 to 70 years (Table 2 , Fig. 1 ). Nevertheless, there appeared to be a myopic shift in those aged 70+ years (Table 2 , Fig. 1 ). A similar (though not statistically significant) relative trend was noted in the men. However, this data in women and men aged 80 years and older should be evaluated with caution, since our sample size in this older age group was relatively small (54 women and 33 men). 

Corneal Power

The data on the mean, SD, and range of CP in Latinos, stratified by gender and age group, are presented in Tables 1 and 2and Figure 2 . The difference between the men and women is significant, both overall and at each age group (P < 0.0001). These gender-related differences in CP were statistically significant after adjusting for height (P = 0.04). 

Figure 2shows the distribution of mean corneal power for each year stratified by gender and the locally weighted regression lines that best fit the data. Overall, for all participants in the 40 to 80+ year age group, each additional year of age correlated with a 0.007-D decrease in corneal power (Table 3) . This difference was statistically significant (P = 0.02) but clinically negligible. Furthermore, when the data was stratified by gender, no significant age-related difference was noted (women, P = 0.48; men, P = 0.06). 

Axial Length

The mean, SD, and range of AL in Latinos, stratified by gender and age group, are presented in Tables 1 and 2 . Overall, in Latinos, the average AL was 23.38 mm, with a range of 18.7 to 33.8 mm in the women (mean, 23.18 mm) and 20.7 to 34.5 mm in males (mean, 23.65 mm). In females, the mean AL ranged from 23.2 mm in those aged 40 to 49 years to 22.9 mm in those aged 80 years and older. In the men, the mean AL was 23.7 mm in those aged 40 to 49 years and 80+ years old (on average, 0.5–0.8 mm longer than in the women from the same age groups; Table 2 ). The difference between the men and women is significant, both overall and at each age group (P < 0.0001). This difference remained significant even after adjusting for height (P < 0.0001). 

Figure 3shows the distribution of mean AL for each year and the locally weighted regression lines that best fit the data. Overall, in all participants from the 40 to 80+ year age group, each year increase in age correlated with a 0.007-mm mean decrease in AL (Table 3) . Similar to the findings for corneal power, this difference was statistically significant (P < 0.0001) but clinically negligible. After stratifying by age group, in the 40- to 49-year age group, each year increase in age correlated with a 0.027-mm mean decrease in AL. However, no significant decrease in AL with age was noted in Latinos 50 years of age and more. After the subjects were stratified by gender, there was no significant age-related difference in AL for either the women or men (women, P = 0.05; men, P = 0.91). The overall difference between the men and women is significant, however, and is demonstrated in the relatively parallel regression lines across age distribution (P < 0.0001; Fig. 3 ). 

Vitreous Chamber Depth

Tables 1 and 2present the data on the mean, SD, and range of VCD in Latinos, stratified by gender and age group. In the men, mean VCD was 15.3 and 15.4 mm in those aged 40 to 49 years and 80+ years old, respectively (on average, 0.3–0.8 mm longer than in the women in the same age groups; Table 2 ). The difference between the men and women is significant, both overall and in each age group. 

The distribution of mean VCD was similar to that of axial length. After the data were stratified by gender, there was no significant age-related difference in VCD in the men (P = 0.89). The older women had significantly shallower VCD when compared with the younger women (P = 0.005). Furthermore, the overall difference between the men and women is significant and is demonstrated by the divergent regression lines when comparing younger and older Latinos across the age distribution (P < 0.0001). These gender-related differences in VCD were statistically significant after adjusting for height (P = 0.03). 

Anterior Chamber Depth

The data on the mean, SD, and range of ACD in Latinos, stratified by gender and age group, are presented in Tables 1 and 2 . The difference between the men and women is significant both overall and at each age group (P < 0.0001). 

Figure 4shows the distribution of mean ACD for each year and the locally weighted regression lines that best fit the data. Other regression lines were explored, but none provided a significantly better fit. Overall, in all participants over the 40 to 80+ year age group, a 1-year increase in age correlated with a mean 0.011-mm decrease in ACD (Table 3) . This difference was statistically significant (P < 0.0001) and probably clinically significant. In addition, after stratifying by age group, in the 40- to 49-, 50- to 59-, and 60- to 69-year age groups, a 1-year increase in age correlated with a mean decrease in ACD of 0.015 mm (P < 0.0001), 0.01 mm (P < 0.0001), and 0.007 mm (P = 0.04), respectively. In Latinos 70 years of age and older, no significant age-related change in ACD was noted. There was a significant age-related difference (P < 0.001) in both the women and men, however. The overall difference between males and females is significant and consistent and is demonstrated in the relatively parallel regression lines across the age distribution (P < 0.0001; Fig. 4 ). These gender-related differences in ACD were statistically significant after adjusting for height (P < 0.0001). 

Lens Thickness

Tables 1 and 2present the mean, SD, and range of lens thickness in Latinos stratified by gender and age group. The difference between the men and women is significant, both overall and at each age group. 

Overall, in all participants in the 40 to 80+ year age group, a 1-year increase in age correlated with a mean 0.01-mm increase in lens thickness (Table 3) . Similar to anterior chamber depth, this difference was both statistically and clinically significant (P < 0.0001). After the data were stratified by age group, in the 40- to 49-year and 50- to 59-year age groups, a 1-year increase in age correlated with a mean increase in lens thickness of 0.014 mm (P < 0.004) and 0.01 mm (P < 0.009), respectively. No significant age-related increase in lens thickness was noted in Latinos 60 years and older. In both the women and men, there was a significant age-related difference (women, P < 0.0001; men, P = 0.0004). In addition, the overall difference between the men and women, while small, is statistically significant (P = 0.007). These gender-related differences in lens thickness were not statistically significant after adjusting for height (P = 0.10). 

Lens Nuclear Opalescence

The frequency distribution for LOCS II NOP grade in Latinos, stratified by gender and age group, are presented in Tables 1 and 2 . In females, NOP was greater in older women with a LOCS II NOP grade of II or higher present in 0.4% aged 40 to 49 years and in 75.8% of those aged 80 years and older. In males, a similar trend was noted with 0.3% of those aged 40 to 49 years and 79.7% of those aged 80 years and older having a LOCS II NOP grade of II or higher (Table 2) . 

Overall, in all participants in the 40 to 80+ year age group, a 1-year increase in age correlated with a 0.04 increase in the LOCS II NOP grade (Table 3) . After the data were stratified by age group, the LOCS II NOP grade is a significantly higher in Latinos from 40 to 79 years of age and marginally significantly higher in those aged 80 years and older (P = 0.07). The overall difference between males and females is not significant (P = 0.15). 

Determinants of Refractive Error

When analyzing the relationship of refractive error to various biometric and clinical parameters, a significant relationship is seen between vitreous chamber depth and refractive error. In addition, significant interrelationships exist between vitreous chamber depth and corneal power and between vitreous chamber depth and lens thickness. We did not put axial length in this model, because axial length is the sum of the vitreous chamber depth, anterior chamber depth, and lens thickness that were included in the model. We also constructed stepwise linear regression models to evaluate the independent determinants of noncycloplegic refractive error—overall (adjusted for age and gender) and stratified by age (adjusted for gender). Two models, both with refractive error as the dependent variable, are presented in Table 4 . In the first model, the independent variables were axial length, corneal power, and lens opalescence. In the second model, the independent variables were anterior chamber depth, lens thickness, vitreous chamber depth, central corneal thickness, corneal power, and lens opalescence. The contributory effect of each independent variable in determining refractive error was estimated by the magnitude of the SRC. This contributory effect was also assessed by the magnitude of the PCC that explained the proportion of variation in refractive error. 

In the overall model (all participants), biometric variables and lens opalescence were significantly associated with refractive error and explained 52% to 54% of the variation in refractive error (model R ²). The most important contributory effect was that of axial length (SRC −0.7, PCC 0.33) or, when axial length was broken down into its components, by vitreous chamber depth (SRC −0.9, PCC 0.27). After axial length and vitreous chamber depth, the next important independent contributory effect to refractive error was by the corneal power of the eye (SRC −0.4, PCC 0.12–0.13). Other variables that also had a small but significant contributory effect on refractive error were lens opalescence (SRC −0.2, PCC 0.03) and the two other components of axial length: lens thickness (SRC −0.4, PCC 0.08) and anterior chamber depth (SRC −0.1, PCC 0.007). 

When the models were stratified by age-group and adjusted for gender, in the 40- to 49-year age group, axial length, and corneal power were significantly associated with refractive error and explained 57% to 58% of the variation in refractive error (model R ²; (Table 4) . The most important contributory effect on refractive error was by axial length (SRC −0.8, PCC 0.44) or, when axial length was broken down into its components, by vitreous chamber depth (SRC −0.9, PCC 0.33). After axial length and vitreous chamber depth, the next important independent contributory effect to refractive error was by the corneal power of the eye (SRC −0.4, PCC 0.12–0.14; Table 4 ). Other variables that also had a small but significant contributory effect on refractive error were the other two components of axial length: lens thickness (SRC −0.4, PCC 0.12) and anterior chamber depth (SRC −0.1, PCC 0.02; Table 4 ). Similar contributions of axial length, corneal power, and lens opalescence were noted in persons aged 50 to 59 years. However, in addition to these biometric variables, in the 50- to 59-year age group, lens opalescence was a small but significant contributor to the magnitude of refractive error (SRC −0.1, PCC 0.02, P < 0.00001; Table 4 ). 

In Latinos aged 60 to 80+ years, the biometric and clinical variables (axial length and its components, corneal power, and lens opalescence) explained a decreasing amount of the variation in refractive error (Table 4) . However, the contributory effect of lens opalescence increased in Latinos aged 40 to 79 years (SRC in 40- to 49-year, 50- to 59-year, 60- to 69-year, and 70- to 79-year age groups were significant: −0.1, −0.2, −0.3, respectively) (Table 4) . Similarly, the variation in refractive error explained by lens opalescence also increased (PCC in 40- to 49-year, 50- to 59-year, 60- to 69-year, and 70- to 79-year age groups were significant, 0.02, 0.04–0.05, and 0.07–0.09, respectively; Table 4 ). Finally, in the 80+ age group, only axial length (and vitreous chamber depth) contributed significantly to the magnitude and variation in refractive error. However, the contribution of these variables was approximately half of that present in those aged 40 to 49 years (Table 4) . 

Discussion

Our study provides the first population-based ocular biometry data for any racial or ethnic group in the US The present study evaluated Latinos, the largest and fastest growing minority group in California and in the United States, comprising 32.4% and 12.5% of these populations, respectively. ¹⁵ In addition to ocular biometric data, we present data on noncycloplegic refraction and lens nuclear opalescence. These cross-sectional data can serve as normative values for Latinos when assessing changes caused by disease or surgical intervention. 

After adjusting for age, we found that men have significantly different ocular biometric, refractive, and clinical measurements compared to women. On average, men are more emmetropic, have less steep corneas (and thus lower corneal power), longer axial lengths, deeper anterior, and vitreous chambers, thicker lenses, and less lens nuclear opalescence. Next, we observed that although there is a hyperopic shift in both female and male Latinos aged 40 to 70 years, a relative myopic shift is present in those 70 years of age and older (Fig. 1) . Other age-related differences include shallower anterior chambers and thicker and more opalescent lenses in older persons compared with younger persons. There was no difference in the axial length or corneal power in older persons compared to younger persons. However, older Latino women had a shallower vitreous chamber compared to younger Latino women. This age-related difference in vitreous chamber depth was not present in men. Finally, axial length (and more specifically, vitreous chamber depth) was the greatest contributor to the magnitude of the noncycloplegic refractive error in the 40 to 80+ year age group. In addition to axial length, the other contributory components in order of importance were corneal power, lens thickness, lens opalescence, and anterior chamber depth. Last, the relative contribution of axial length decreased and that of lens opalescence increased in older persons compared to younger persons. 

The initial age-related hyperopic shift followed by a myopic shift in noncycloplegic refractive error have also been noted in other population-based studies in whites in the Beaver Dam, ¹⁶ Blue Mountains, ¹⁷ and Melbourne ¹⁸ studies, in Afro-Caribbeans in the Barbados study ¹⁹ , in both whites and African-Americans in the East Baltimore study, ²⁰ in Mongolians, ¹¹ and in the Singaporean Chinese. ¹⁰ Although the exact reason for the hyperopic shift in older persons is unclear, it may be due to the loss of accommodation in older persons compared to younger persons. Because we used noncycloplegic refraction any underlying hyperopia in younger persons would have been masked by the accommodative drive, whereas the loss of accommodative ability that comes with age would have unmasked the hyperopia in older persons. In contrast, the myopic shift in older Latinos (70+ years of age) may be due to the greater degree of lens opalescence. In support of this hypothesis, we found that after age-adjustment, participants with significant lens opalescence were significantly more myopic than those without lens opalescence. 

To date, few population-based studies have reported biometric and related clinical data to describe age- and gender-related differences and assess the contribution of various components of the eye to refractive error in adults. ^{10 11} The two previous studies that have presented such detailed population-based data are from East Asia –the Mongolian Eye study ¹¹ and the Tanjong Pagar Survey from Singapore. ¹⁰ The pattern of noncycloplegic refractive error is consistent between these studies and our study. In our study cohort, however, the hyperopic shift appears to occur up to a decade later (persons aged 40–70 years) than that observed in either Mongolia or Singapore (persons aged 40–60 years). Also, in our cohort, those persons aged 70 years and older were noted to have a relative myopic shift; again, this is a decade later than that observed in Mongolian and Singaporean Chinese. The reason for these differences is unclear. 

The specific determinants of refractive error in our study were primarily axial length (and more specifically vitreous chamber depth), corneal power, and lens opalescence. The contribution of corneal power remains consistent over the 40 to 80+ year age group. However, the relative contribution of axial length is smaller and that of lens opalescence is greater in older persons compared to younger persons. These differences in lens opalescence are probably related to changes in the refractive index of the lens noted in laboratory studies of lenticular aging. ²¹ Although this general pattern is consistent with that seen in Singaporean Chinese, our findings differ in significant ways. In our study, lens nuclear opalescence was an important contributor to refractive error in the overall model, and more specifically in those aged 50 years and older (Table 4) . In the Chinese, lens opalescence was significant and important in those aged 60 years and older but not in those aged 40–59 years. ¹⁰ Further supporting our observation is the fact that in our study the contributions of lens opalescence were paralleled by the contributions of lens thickness in those aged 50 years and older. Similar to our study, in the Chinese, lens thickness was an important and significant contributor in all persons, both younger (40–59 years) and older (60–81 years). ¹⁰ One explanation for these differences may be that, in the Chinese study, persons were divided into only two groups (40–59 and 60–81 years) whereas in our study we grouped all persons by decade (40–49, 50–59, 60–69, and so forth). These smaller groups allowed us to obtain a narrower range of ages so that we could better assess age-related differences. 

In terms of biometric measurements, the two previously published population-based studies have found contradictory age-related differences. ^{10 11} In Singapore, younger adult Chinese had longer axial lengths compared to older Chinese. ¹⁰ In contrast, in Mongolia, there was no age-related difference in axial length. ¹¹ Similar to the Mongolians, in our study, no age-related differences in axial length were noted. One explanation for this is that once the eye has achieved its adult size, little change occurs in the axial length during adulthood and with aging. The observation in Singaporean Chinese of shorter axial length in older persons may be attributable to a cohort effect. It is likely that Chinese who were born more recently may perform more near-work activities during the ocular growth phase compared to older Chinese or there may be a differential in other environmental factors that may be related to longer axial lengths. When comparing age-related differences in vitreous chamber depth, we did note a small but significant difference in females, similar to that seen in Chinese females in Singapore. However, in the men in our study we were unable to find a difference, similar to findings in the Mongolians. We are unable to explain this gender-related difference in vitreous chamber depth in our population. Whereas these differences do exist, there are significant similarities between the three studies. In all three studies, older persons had shallower anterior chambers, greater lens thickness, and no difference in corneal power compared to younger persons. These changes can be explained by aging of the lens causing the thicker lenses in older persons. ²¹ The thicker lenses may then shift the iris forward, leading to shallower anterior chambers. This shallowing of the anterior chamber was accompanied by concurrent thickening and anterior displacement of the lens, as has been noted in other studies. ^{6 9 10 11 22 23 24 25 26 27} The age-related shallowing of the anterior chamber may help explain the increased prevalence of primary angle-closure glaucoma in older individuals. ^{22 23 25 26 27} Further, the increased risk for women developing this condition may be due to their having shallower anterior chambers than men at all age groups. 

Finally, a higher degree of lens opalescence noted in older persons in our study has been noted in other population-based studies. ^{28 29 30 31 32 33} In addition, the higher grades of lens nuclear opalescence have also been previously reported in various population-based studies. ^{28 29 31} These data on higher grades of lens opalescence in older persons also support other laboratory studies demonstrating oxidative damage and opacification of the human lens with aging. ²¹ Thus, this observation is not likely due to a cohort effect. 

Although the relative pattern of age-related differences in the three studies has been discussed above, Singaporean Chinese ¹⁰ (women SE −2.08 to −0.42 D) are on average one diopter more myopic across the 40 to 80+ years age spectrum compared to Latinos (women SE −0.32 to +0.74 D). In contrast, the Latinos in our study had similar noncycloplegic refractive error to Mongolians (women SE −0.3 to +0.4 D). ¹¹ In addition, no systematic differences were noted in the axial length of the three populations. However, there were systematic differences in the vitreous chamber depth and anterior chamber depth, with the anterior chamber being shallower and the vitreous chamber deeper in Chinese (e.g., females anterior chamber depth ranges from 2.55 to 3.08 mm; vitreous chamber depth ranges from 14.98 to 15.88 mm) compared to Latinos (e.g., the women’s anterior chamber depth ranges from 3.17 to 3.49 mm; vitreous chamber depth ranges from 14.7 to 14.98 mm) in all age groups. ¹⁰ These differences in biometric measurements were not present when comparing Mongolians and Latinos. Although potential explanations for these differences may be more likely to be related to differences in near-work activity than a genetic explanation, the causes of these differences remain unclear. 

Our study has several strengths. First, it provides the first population-based data on ocular biometry, lens opalescence, and noncycloplegic refraction in a United States population. To our knowledge, it is the first population-based study of Latinos to provide such data. One strength of our study was our success in obtaining ocular biometry and lens opalescence data on 72% of the total eligible Latino cohort. Population-based data are important, because such data are unlikely to have selection biases that could be present in smaller clinic-based data sets. Further, our sample size of 5588 participants with biometric data—a significantly larger sample size than that of the previous population-based studies—allowed us to evaluate age-related differences carefully, especially in older persons. Another strength of our study is the use of standardized protocols for obtaining biometric measurements and refractive error. Such standardized protocols allows comparison of our data to other population-based data. 

The cross-sectional nature of our study is one limitation in assessing aging changes in refractive, biometric, and clinical characteristics. Although inferences can be drawn about differences in biometric measurements in younger and older individuals, these differences need to be interpreted with caution, since these differences may reflect a cohort effect rather than a longitudinal age-related change. Another limitation of our study is that we used noncycloplegic refraction. Because noncycloplegic refraction does not account for accommodative power, it is likely that some of the hyperopic shift noted in older persons compared to younger persons is a reflection of the loss of accommodation in older persons rather than a true hyperopic shift. Finally, although the study subjects who were included were younger, less likely to be born outside the United States, less likely to be unemployed, and had a higher level of acculturation, and thus may have an overall higher prevalence of myopia, it is unlikely to have any impact on the relationship between refractive error and the various biometric and clinical variables. 

In summary, older persons in the present study had shorter axial and vitreous chamber lengths, shallower anterior chambers, and thicker lenses than younger individuals. These differences in biometry were associated with a trend toward greater hyperopic refractive errors until the age of 70 years when a myopic shift was observed; this shift was strongly associated with the presence of nuclear opalescence. The strongest determinants of refractive error across the age spectrum were axial length (specifically vitreous chamber depth) and corneal curvature/power. NOP contributes a small but significant amount to the determination of refractive error, particularly in older Latinos. Finally, a majority of the variations in refractive error in older Latinos (70 years and older) is unexplained by either biometric parameters or lens opalescence, and deserves further study. 

Appendix 1

The Los Angeles Latino Eye Study Group, University of Southern California, Los Angeles, California: Rohit Varma, Sylvia H. Paz, LaVina Abbott; Stanley P. Azen, Lupe Cisneros, Elizabeth Corona, Carolina Cuestas, Denise R. Globe, Sora Hahn, Mei Lai, George Martinez, Susan Preston-Martin, Ronald E. Smith, Mina Torres, Natalia Uribe, Jennifer Wong, Joanne Wu, and Myrna Zuniga. 

Battelle Survey Research Center, St. Louis, Missouri: Sonia Chico, Lisa John, Michael Preciado, and Karen Tucker. 

Ocular Epidemiology Grading Center, University of Wisconsin, Madison, Wisconsin: Ronald Klein. 

 Supported by Grants EY11753 and EY03040 from the National Eye Institute and the National Center on Minority Health and Health Disparities, National Institutes of Health, Bethesda, Maryland, and an unrestricted grant from Research to Prevent Blindness. RV is a Research to Prevent Blindness Sybil B. Harrington Scholar.

 Submitted for publication April 5, 2005; revised August 11, 2005; accepted October 12, 2005.

 Disclosure: C. Shufelt, None; S. Fraser-Bell, None; M. Ying-Lai, None; M. Torres, None; R. Varm a, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Ocular Biometry, Lens Opalescence, and Noncycloplegic Refraction in the LALES: Overall and Stratified by Gender

Variables	All (n = 5588)	Men (n = 2337)	Women (n = 3251)	P
Spherical equivalent (D)	0.11 ± 1.89	0.02 ± 1.66	0.18 ± 2.04	0.001
Corneal power (D), †	43.72 ± 1.62	43.35 ± 1.64	43.95 ± 1.6	<0.0001
Axial length (mm)	23.38 ± 1.01	23.65 ± 0.94	23.18 ± 1.02	<0.0001
Vitreous chamber depth (mm)	15.04 ± 1.06	15.22 ± 1.03	14.91 ± 1.06	<0.0001
Anterior chamber depth (mm)	3.41 ± 0.35	3.48 ± 0.34	3.36 ± 0.34	<0.0001
Lens thickness (mm)	4.38 ± 0.60	4.40 ± 0.64	4.36 ± 0.56	0.007
Lens nuclear opalescence (LOCS II grade), † n (%), ‡
NO	1890 (33.8)	806 (34.5)	1084 (33.3)	0.15
NI	3253 (58.2)	1364 (58.4)	1889 (58.1)
NII	341 (6.1)	125 (5.4)	216 (6.6)
NIII	82 (1.5)	30 (1.3)	52 (1.6)
NIV	22 (0.4)	12 (0.5)	10 (0.3)

 Data are expressed as the Means ± standard deviation.

*  Differences in means between the men and women, age adjusted; and χ² test for lens nuclear opalescence.

†  Sample size is: All n = 2102; men n = 822; women n = 1280.

‡  Lens nuclear opalescence: NO (≤N0); NI (>N0 ≤NI); NII (>NI ≤NII); NIII (>NII ≤NIII); NIV (>NIII). Frequencies and percent of total number for All, Men and Women.

Table 2.

 

Ocular Biometry, Lens Opalescence, and Noncycloplegic Refraction in the LALES: Stratified by Age and Gender

Variable	Women											Men
Age (y)	40–49	50–59	60–69	70–79	80+		40–49	50–59	60–69	70–79	80+
n	1304	1009	631	253	54		923	726	441	214	33
Mean	44	54	64	74	83		45	54	64	74	83
SD	2.8	2.9	2.8	2.7	3.8	P *	2.8	2.9	2.9	2.8	2.8	P *
Axial length (mm)
Mean ± SD	23.2 ± 1.1	23.2 ± 0.9	23.1 ± 1.0	23.1 ± 0.9	22.9 ± 0.9	0.05	23.7 ± 1.0	23.6 ± 0.8	23.6 ± 0.9	23.5 ± 0.9	23.7 ± 0.9	0.91
Range	18.7–33.8	20.4–29.4	19.4–30.3	20.1–27.9	21.3–25.2		20.7–34.5	20–28	21–28	21–28	22–27
Anterior chamber depth (mm)						<0.0001						<0.0001
Mean ± SD	3.49 ± 0.3	3.34 ± 0.3	3.24 ± 0.3	3.17 ± 0.3	3.09 ± 0.3		3.61 ± 0.31	3.47 ± 0.31	3.37 ± 0.3	3.26 ± 0.34	3.3 ± 0.4
Range	2.39–4.43	2.04–4.41	2.16–4.20	2.41–3.98	2.53–3.88		2.49–4.68	2.45–4.45	2.28–4.28	2.32–4.20	2.3–4.08
Lens thickness (mm)
Mean ± SD	4.22 ± 0.63	4.37 ± 0.46	4.49 ± 0.50	4.63 ± 0.51	4.72 ± 0.47	<0.0001	4.2 ± 0.7	4.4 ± 0.6	4.6 ± 0.6	4.7 ± 0.6	4.6 ± 0.4	0.0004
Range	2.56–8.23	3.37–8.03	2.04–8.22	2.78–8.55	3.73–6.22		2.86–8.20	3.04–8.35	2.3–7.9	2.0–8.0	3.5–5.4
Vitreous chamber depth (mm)
Mean ± SD	14.98 ± 1.2	14.92 ± 1.0	14.8 ± 1.03	14.7 ± 1.02	14.6 ± 0.88	0.005	15.3 ± 1.1	15.2 ± .9	15 ± 0.9	15.0 ± 1.04	15.4 ± 0.9	0.89
Range	10.6–25.7	11.7–20.6	10.7–21.5	10.9–19.8	12.6–16.4		10.9–26.5	11.3–20.5	11.0–18.4	10.7–18.9	13.9–17.6
Spherical equivalent (D)
Mean ± SD	−0.32 ± 1.8	0.23 ± 1.9	0.73 ± 2.36	1.02 ± 2.30	0.74 ± 2.12	<0.0001	−0.3 ± 1.3	0.02 ± 1.6	0.4 ± 1.6	0.6 ± 2.5	−0.3 ± 2.6	0.71
Range	−18.5–10.6	−19.0–7.0	−19.0–7.8	−9.3–7.9	−3.75–6.13		−14.0–7.25	−14.3–5.3	−11–5.8	−12.4–9.8	−5.9–4.4
Corneal power (D)
Mean ± SD	44.14 ± 1.6	43.87 ± 1.5	43.9 ± 1.5	43.9 ± 1.6	44.2 ± 1.53	0.48	43.6 ± 1.9	43.4 ± 1.6	43.3 ± 1.4	43.3 ± 1.5	42.8 ± 1.8	0.06
Range	39.0–51.9	39.5–48.1	38.9–48.4	38.0–49.3	40.6–47.0		34.4–52.8	39.1–49.8	40–48	39.9–48.3	38.6–46.4
Lens opalescence (LOCS II Grade) n (%), †
n	923	726	441	214	33		1304	1009	631	253	54
NO	586 (63.5)	184 (25.3)	32 (7.3)	4 (2)	0 (0)	<0.0001	797 (61.1)	240 (23.8)	42 (6.7)	5 (1.9)	0	<0.0001
NI	333 (36.1)	533 (73.4)	361 (81.9)	129 (60.3)	8 (24.2)		504 (38.7)	752 (74.5)	487 (77.1)	135 (53.4)	11 (20.4)
NII	2 (0.2)	7 (0.9)	40 (9.0)	60 (28.0)	16 (48.5)		2 (0.2)	15 (1.5)	88 (13.9)	86 (33.9)	25 (46.3)
NIII	1 (0.1)	1 (0.1)	5 (1.1)	16 (7.5)	7 (21.2)		1 (0.1)	2 (0.25)	11 (1.7)	23 (9.1)	15 (27.8)
NIV	1 (0.1)	1 (0.1)	3 (0.7)	5 (2.3)	2 (6.1)		0 (0)	0 (0)	3 (0.5)	4 (1.6)	3 (5.6)

 Corneal power sample size for each age group is: Women: 40–49 = 352; 50–59 = 366; 60–69 = 342; 70–79 = 178; ≥80 = 42; men: 40–49 = 204; 50–59 = 235; 60–69 = 218; 70–79 = 138; ≥80 = 27.

*  Test for trend by age groups.

†  Lens opalescence grades are as defined in Table 1 . Percentages are of total number in the age group.

Figure 1.

 

Age-specific mean noncycloplegic refraction (spherical equivalent in diopters) by age and stratified by gender with the locally weighted regression line for participants in the Los Angeles Latino Eye Study (LALES). For all figures, the age-specific data from the right eye of all persons by each year of age were plotted (n = 5588) to show the relationship of the variable to age.

Table 3.

 

Linear Regression Models of Ocular Biometric and Clinical Variables Overall and Stratified by Age Groups, in the LALES

Biometric and Clinical Variables	Regression Coefficient	P
All Participants
Spherical equivalent (D)	0.04	<0.0001
Axial length (mm)	−0.007	<0.0001
Corneal power (D)	−0.007	0.02
Vitreous chamber depth (mm)	−0.009	<0.0001
Anterior chamber (mm)	−0.011	<0.0001
Lens thickness (mm)	0.01	<0.0001
Lens nuclear opalescence (LOCS II grade)	0.04	<0.0001
40–49 age group
Spherical equivalent (D)	0.048	<0.0001
Axial length (mm)	−0.027	0.0004
Corneal power (D)	0.035	0.17
Vitreous chamber depth (mm)	−0.027	0.002
Anterior chamber (mm)	−0.015	<0.0001
Lens thickness (mm)	0.014	0.004
Lens nuclear opalescence (LOCS II grade)	0.06	<0.001
50–59 age group
Spherical equivalent (D)	0.07	<0.0001
Axial length (mm)	−0.014	0.07
Corneal power (D)	0.004	0.84
Vitreous chamber depth (mm)	−0.011	0.16
Anterior chamber (mm)	−0.01	<0.0001
Lens thickness (mm)	0.01	0.009
Lens nuclear opalescence (LOCS II grade)	0.03	<0.0001
60–69 age group
Spherical equivalent (D)	0.02	0.47
Axial length (mm)	0.005	0.62
Corneal power (D)	0.013	0.54
Vitreous chamber depth (mm)	0.004	0.69
Anterior chamber (mm)	−0.007	0.04
Lens thickness (mm)	0.009	0.13
Lens nuclear opalescence (LOCS II grade)	0.04	<0.0001
70–79 age group
Spherical equivalent (D)	0.001	0.98
Axial length (mm)	0.01	0.43
Corneal power (D)	−0.007	0.84
Vitreous chamber depth (mm)	0.03	0.13
Anterior chamber (mm)	−0.002	0.76
Lens thickness (mm)	−0.01	0.18
Lens nuclear opalescence (LOCS II grade)	0.07	<0.0001
80+ age group
Spherical equivalent (D)	−0.07	0.30
Axial length (mm)	0.04	0.22
Corneal power (D)	0.04	0.49
Vitreous chamber depth (mm)	0.035	0.20
Anterior chamber (mm)	0.0003	0.98
Lens thickness (mm)	0.00009	0.99
Lens nuclear opalescence (LOCS II grade)	0.047	0.07

 In each regression model, the biometric and clinical variables were the dependent variables, and age was the independent variable. All data reflect adjustment for gender. For example, after adjustment for gender, each 1-year increase in age in the 40–49-year age range, would correlate with a 0.027-mm decrease in axial length, and a 0.015-mm decrease in anterior chamber depth. Regression coefficients in bold are statistically significant.

Figure 2.

 

Age-specific mean corneal curvature (diopters) by age and stratified by gender with the locally weighted regression line for participants in the Los Angeles Latino Eye Study (LALES).

Figure 3.

 

Age-specific mean axial length (in millimeters) by age and stratified by gender with the locally weighted regression line for participants in the Los Angeles Latino Eye Study (LALES).

Figure 4.

 

Age-specific mean anterior chamber depth (in millimeters) by age and stratified by gender with the locally weighted regression line for participants in the Los Angeles Latino Eye Study (LALES).

Table 4.

 

Stepwise Regression Models for Noncycloplegic Refraction (Dependent Variable) and Biometric and Clinical Variables (Independent Variables) for All Participants Stratified by Age in the LALES

Model 1									Model 2
Biometric and Clinical Variables	Regression Coefficient (95% CI)	SRC	PCC	P	Biometric and Clinical Variables	Regression Coefficient (95% CI)	SRC	PCC	P
All participants
Axial length	−1.6 (−1.7 to −1.5)	−0.7	0.33	<0.0001	Vitreous chamber depth	−1.8 (−1.9 to −1.8)	−0.9	0.27	<0.0001
					Corneal power	−0.6 (−0.7 to −0.6)	−0.4	0.12	<0.0001
Corneal power	−0.6 (−0.6 to −0.5)	−0.4	0.13	<0.0001	Lens thickness	−1.5 (−1.7 to −1.4)	−0.4	0.08	<0.0001
Nuclear opalescence	−0.7 (−0.8 to −0.6)	−0.2	0.03	<0.0001	Nuclear opalescence	−0.7 (−0.8 to −0.6)	−0.2	0.03	<0.0001
					Anterior chamber	−0.6 (−0.8 to −0.4)	−0.1	0.007	<0.0001
Model R 2			0.52		Model R 2			0.54
40–49 age group
Axial length	−1.6 (−1.7 to −1.5)	−0.8	0.44	<0.0001	Vitreous chamber depth	−1.7 (−1.9 to −1.6)	−0.9	0.33	<0.0001
Corneal power	−0.5 (−0.6 to −0.5)	−0.4	0.14	<0.0001	Corneal power	−0.6 (−0.7 to −0.5)	−0.4	0.12	<0.0001
					Lens thickness	−1.6 (−1.8 to −1.3)	−0.4	0.12	<0.0001
					Anterior chamber	−1.0 (−1.4 to −0.6)	−0.1	0.02	<0.0001
Model R 2			0.57		Model R 2			0.58
50–59 age group
Axial length	−1.8 (−1.9 to −1.6)	−0.8	0.41	<0.0001	Vitreous chamber depth	−2.1 (−2.2 to −1.9)	−0.9	0.35	<0.0001
Corneal power	−0.6 (−0.7 to −0.5)	−0.4	0.15	<0.0001	Corneal power	−0.7 (−0.8 to −0.6)	−0.5	0.13	<0.0001
Nuclear opalescence	−0.6 (−0.8 to −0.4)	−0.1	0.02	<0.0001	Lens thickness	−1.8 (−2.0 to −1.5)	−0.4	0.12	<0.0001
					Nuclear opalescence	−0.6 (−0.9 to −0.4)	−0.1	0.02	<0.0001
					Anterior chamber	−0.5 (−0.8 to −0.1)	−0.07	0.004	0.01
Model R 2			0.58		Model R 2			0.62
60–69 age group
					Vitreous chamber depth	−1.8 (−1.9 to −1.6)	−0.8	0.27	<0.0001
Axial length	−1.5 (−1.7 to −1.4)	−0.7	0.28	<0.0001	Corneal power	−0.7 (−0.8 to −0.6)	−0.5	0.14	<0.0001
Corneal power	−0.7 (−0.8 to −0.6)	−0.4	0.16	<0.0001	Lens thickness	−1.3 (−1.6 to −1.0)	−0.3	0.06	<0.0001
Nuclear opalescence	−0.8 (−1.1 to −0.6)	−0.2	0.05	<0.0001	Nuclear opalescence	−0.8 (−1.0 to −0.5)	−0.2	0.04	<0.0001
					Anterior chamber	−0.5 (−0.9 to −0.1)	−0.08	0.006	0.01
Model R 2			0.49		Model R 2			0.52
70–79 age group
Axial length	−1.6 (−1.9 to −1.4)	−0.6	0.20	<0.0001	Vitreous chamber depth	−1.9 (−2.2 to −1.6)	−0.8	0.18	<0.0001
Corneal power	−0.6 (−0.7 to −0.4)	−0.4	0.10	<0.0001	Corneal power	−0.6 (−0.8 to −0.5)	−0.4	0.09	<0.0001
Nuclear opalescence	−0.9 (−1.2 to −0.7)	−0.3	0.09	<0.0001	Lens thickness	−1.5 (−1.9 to −1.0)	−0.4	0.07	<0.0001
					Nuclear opalescence	−0.9 (−1.1 to −0.6)	−0.3	0.07	<0.0001
Model R 2			0.40		Model R 2			0.42
80+ age group
Axial length	−0.9 (−1.5 to −0.3)	−0.4	0.12	0.002	Vitreous chamber depth	−1.0 (−1.6 to −0.4)	−0.4	0.15	0.0008
Model R 2			0.20		Model R 2			0.22

 In the model for all participants, we adjusted for both age and gender. For the models stratified by age, we adjusted for gender. Model 1: Variables included in the model were axial length, corneal power, central corneal thickness, and nuclear opalescence. For example, increase in axial length corresponds to a −1.6 D shift in refractive error in all participants. Model 2: Variables included in the model were: vitreous chamber depth, corneal power, lens thickness, anterior chamber depth, central corneal thickness, and nuclear opalescence. In this model, axial length has been broken down into its components: vitreous chamber depth, anterior chamber depth, lens thickness, and central corneal thickness. For example, a 1-mm increase in vitreous chamber depth corresponds to a −1.8 D shift in refractive error in all participants. Model R² is cumulative for the model. SRC, standardized regression coefficient; PCC, partial correlation coefficient.

The authors thank the LALES External Advisory Committee for their advice and contributions: Roy Beck, (Chairman); Natalie Kurinij, Leon Ellwein, Helen Hazuda, Eve Higginbotham, Lee Jampol, M. Cristina Leske, Donald Patrick, and James M. Tielsch.