The interdisciplinary Gutenberg Health Study (GHS) is a prospective, population-based, observational cohort study in the Rhine-Main region of Western Germany (Rhineland-Palatinate).20 Recruitment and baseline examination were performed between 2007 and 2012 of participants aged between 35 and 74 years. The initial study design contained consecutive follow-ups every 5 years. Study participants were randomly drawn stratified by sex and residence (urban or rural) from local governmental registry offices where every resident has to be registered. For the present analysis, participants of the 5-year follow-up conducted between 2012 and 2017 were included because in the follow-up examination, every participant was examined with Scheimpflug imaging. Overall, 12,423 of 15,010 GHS participants (82.8%) returned for 5-year follow-up examination. In each participant, a detailed assessment of medical history was performed by a personal interview and an extensive ophthalmological examination was conducted. Written informed consent was obtained from all study participants prior to their entry into the study and the GHS complies with Good Clinical Practice (GCP), Good Epidemiological Practice (GEP), and the ethical principles of the Declaration of Helsinki. The study protocol and study documents were approved by the local ethics committee of the Medical Chamber of Rhineland-Palatinate, Germany (reference no. 837.020.07; original vote: 22.3.2007, latest update: 20.10.2015).

Participants were asked upon invitation to the GHS study center to review their records or family albums for a documented birth weight and were divided into the following birth weight groups: a low birth weight group < 2500 g (group 1); a normal birth weight group with a birth weight between 2500 and 4000 g (group 2); and a high birth weight group > 4000 g (group 3) as previously reported.21-25 Additionally, participants with birth weights below 1000 g and above 6000 g were excluded as these self-reported data are suspected to be unreliable. The distribution of low, normal and high birth weight in our cohort was compared to reported birth weight data in medical literature and to data of the German Federal Statistical Office as reported earlier.21,26,27 Furthermore, we validated birth weight by reviewing for a subsample of participants born in the UMCM documented birth weight in the birth register.

The ophthalmic examination of the GHS was described earlier.28 In brief, ocular examination consisted of testing distant corrected visual acuity and objective refraction (Humphrey® Automated refractor/Keratometer (HARK) 599™). Distant corrected visual acuity was measured using the built-in Snellen charts, ranging from 20/400 to 40/20 (decimal 0.05 to 2.0). Intraocular pressure was measured with a noncontact tonometer (NT 2000™, Nidek Co./Japan). Imaging of the macula and optic nerve head was conducted with a non-mydriatic fundus camera.28 Scheimpflug imaging (Pentacam™, Oculus, Wetzlar, Germany) and optical biometry (Lenstar LS900, Haag-Streit, Bern, Switzerland) was performed.28 The spherical equivalent was calculated by adding the spherical correction value to half the cylinder value.

The anterior segment examination was performed with a Pentacam HR rotating Scheimpflug camera. Scheimpflug imaging enables anterior segment measurements from the anterior cornea surface to the posterior lens surface. All examinations were performed according to standardized operating procedures to diminish examiner dependent variance. Corneal tomography was based on 25 Scheimpflug images recorded within two seconds. The following corneal aberrations (analysis including anterior and posterior surface of the cornea) were recorded: oblique and vertical astigmatism (Z2–2; Z22), corneal vertical and horizontal coma (Z3–1; Z31), corneal vertical and horizontal trefoil (Z3–3; Z33), corneal spherical aberration (Z40) and corneal root-mean square of higher-order aberrations (RMS HOA; 3rd up to 8th order) and of lower-order aberrations (RMS LOA). Lower order aberrations included 1st and 2nd order aberrations (tilt, astigmatism, defocus). Zernike parameters were calculated on an optic zone of 6 mm diameter. A refractive index of 1.3375 was used to adjust the measurements of the anterior surface to match keratometer results. As reported by the manufacturer, ray tracing is used to calculate aberrations of the total cornea using the wavefront of light rays passing parallel through the cornea and to indicate the differences from an ideal (spherical) wavefront using both the shape of the anterior and posterior corneal surface.

Validation of all measurements was performed by checking quality description of the device. Furthermore, the graphical distribution of corneal parameters was analyzed and skewness was calculated. In the case of skewness ≤ |1|, we assumed approximately normal distribution. Scheimpflug images with measurements above the 4th standard deviation were manually graded by two board-certified ophthalmologists. If parts of the Scheimpflug images were of unsatisfied quality the complete Scheimpflug measurement of the participant was excluded. In every participant only one Scheimpflug measurement was performed.

Factors that may affect corneal aberrations in our participants were considered as covariates. These included (1) age (in years), (2) sex (reference: male), (3) corneal power (in diopter), (4) axial length (mm), (5) white-to-white (mm), (6) pseudophakia (yes), (7) intraocular pressure (mmHg).

For the present analysis, only participants with available self-reported birth weight, no reported history of eye injury, no report of previous corneal surgery and sufficient Scheimpflug imaging were included.

A detailed statistical analysis plan was developed before conducting the analysis. The primary outcomes were total corneal aberrations as well as corneal HOA and LOA aberrations. As secondary outcome the association of birth weight with oblique and vertical astigmatism (Z2−2; Z22), corneal vertical and horizontal coma (Z3−1; Z31), corneal vertical and horizontal trefoil (Z3−3; Z33) and corneal spherical aberration (Z40) were assessed. A comparison of characteristics of included and excluded participants was conducted. Descriptive statistics were computed. Absolute and relative frequencies were calculated for dichotomous parameters, the mean and standard deviation were calculated for approximately normally distributed data, and the median and interquartile range were calculated for the remaining variables. For an eye level analysis linear regression models with generalized estimating equations (GEE) were used to assess associations and to account for correlations between corresponding eyes. Before conducting statistical analysis, the terms astigmatism oblique, coma horizontal and trefoil horizontal of the left eyes were transformed (multiplied with −1), as they showed an inverse correlation with the right eyes. In model #1, self-reported birth weight was the only predictor for the main outcome measures; in model #2, adjustment for age (in years), sex (female), corneal power (in diopter), axial length (mm), white-to-white (mm), pseudophakia (yes) and IOP (mmHg) was performed. As this is an explorative study, P values should be regarded as a continuous parameter reflecting the level of evidence; and are therefore reported exactly.29 A p-value <0.001 was considered as strong association and p<.05 as likely association and p≥.05 as not significant. R2 was calculated by the change when including birth weight as additional parameter in the multivariable model. Furthermore, two sensitivity analyses were performed. In the first analysis, participants wearing contact lenses were excluded, and in the second analyses pseudophagic subjects were excluded. All models were computed in a first step including birth weight as a continuous variable and in a second step including birth weight as a categorical variable (birth weight < 2500 g, birth weight between 2500 and 4000 g, a birth weight > 4000 g). The data were analyzed with R version 3.6.1 (2019–07–05) R Core Team (2019). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Of the 15,010 participants at baseline 12,423 were examined at the five-year follow-up. Overall, of these participants 5269 participants were excluded because of missing birth weight data or a birth weight above 6000 g or below 1000 g. Furthermore, in 1002 participants Scheimpflug measurement was not available or not sufficient, 121 were excluded because of self-reported corneal surgery or missing information about self-reported corneal surgery and 403 because of self-reported corneal injury or missing information. In total, 5628 were included in the present study with available self-reported birth weight and Scheimpflug measurements. The mean age of study participants was 56.0 ± 10.3 years and 53.4% (n = 3004) were female. Birth weight below 2500 g (group 1) was reported by 313 participants, birth weight between 2500 g and 4000 g (group 2) was reported by 4629 participants, and a birth weight above 4000 g (group 3) was reported by 686 participants (Table 1). Measurements of corneal aberrations by birth weight groups with Scheimpflug tomography are displayed in Table 2. In Fig. 1 scatterplots of birth weight with spherical aberrations and higher-order aberrations are presented.

Fig. 1.

Scatterplot of birth weight with a) spherical aberrations and b) higher-order aberrations in the Gutenberg Health Study (n = 5628). Participants with lower birth weights showed higher spherical aberration and increased higher-order aberrations.

(0.3MB).

In the non-responder analysis included subjects were younger (56.0 ± 10.3 years versus 62.4 ± 10.5 years) and more likely female (53.4% versus 45.0%).

Lower birth weight was strongly associated with increased spherical aberrations in multivariable model #2 (per 500 g: B=−0.006 µm [95%-CI: −0.008; −0.003], p<.001; R2 = 0.003). A strong association was also observed for lower birth weight and increased higher order aberrations in model #2 (per 500 g: B=−0.007 µm [95%-CI: −0.010; −0.003], p<.001; R2 = 0.002) and increased lower order aberrations in model #2 (per 500 g: B=−0.027 µm [95%-CI: −0.041; −0.013], p<.001; R2 = 0.002) (Fig. 2). In multivariable model #2 a likely association was observed between birth weight and horizontal trefoil (Z33), while no statistic significant association was observed between birth weight and oblique (Z2−2) and vertical astigmatism (Z22), horizontal (Z31) and vertical coma (Z3−1), and vertical trefoil (Z3−3) (Fig. 2). Corneal aberrations for corneal front and back surface are presented in appendix Table 1 and 2.

Fig. 2.

Associations of the different corneal aberration types with birth weight (continuous) in the Gutenberg Health Study (2012–2017).

Linear regression analysis using generalized estimating equations to control for correlations between right and left eyes. Estimates are present per 500 g. B – Beta; CI – Confidence interval

M1 Crude model without adjustment.

M2 Model adjusted for age; sex; corneal power, axial length, white-to-white, pseudophakia, IOP.

A p-value <0.001 was considered as strong association, p<.05 as likely association, p≥.05 as not statistically significant association.

(0.23MB).

When analyzing associations with birth weight categorized into low, normal and high birth weight, no strong statistically significant associations were observed between low birth weight and the different types of aberrations. A likely association was observed between low birth weight and decreased horizontal trefoil (Z33). The higher birth weight group revealed in comparison to the normal birth weight group in the multivariable analysis after adjustment for potential confounders a strong association with lower spherical aberrations (Z40), a likely association with lower horizontal coma (Z31) and decreased total corneal and lower order aberrations while no statistically significant association was observed with the other different types of aberrations (Fig. 3).

Fig. 3.

Associations of the different corneal aberration types with Birth Weight Groups (low birth weight n = 313; normal birth weight n = 4629; high birth weight n = 686) in the Gutenberg Health Study (2012–2017).

B – Beta; CI – Confidence interval

Linear regression analysis using generalized estimating equations to control for correlations between right and left eyes.

M1 Crude model without adjustment.

M2 Model adjusted for age; sex; corneal power, axial length, white-to-white, pseudophakia, IOP.

A p-value <0.001 was considered as strong association, p<.05 as likely association, p≥.05 as not statistically significant association.

(0.34MB).

After exclusion of persons wearing contact lenses, the association analysis revealed comparable results. Furthermore, the exclusion of pseudophagic persons the results remained stable.

One of the major limitations of the present study is that 42.4% of all GHS study participants did not provide self-reported birth weight data at baseline. Included participants were more frequently female and younger. Our data are additionally restricted because self-reported birth weight data were not validated by review of medical records for all participants. In order to ensure a high validity of self-reported birth weight data in our analyses we asked every study participant at study invitation to review personal records or family albums for recorded birth weight data and previous authors found a high reliability of self-reported birth weight10 thus we assume that this is a valid parameter in our study. In addition, we compared self-reported birth weight of a subsample of participants (n = 86) born in the UMCM. We observed a high correlation between self-reported birth weight and documented birth weight in the birth register (Correlation coefficient: 0.89 95%-CI 0.83; 0.92). Furthermore, birth weight distribution of our cohort was compared to previous reports of the German Federal Statistical Office reporting comparable distributions as reported earlier.21,42 Therefore, we assume that our findings can be generalized to the German population. Other limitations are missing perinatal data such as gestational age, intrauterine growth restriction, postnatal occurrence of ROP. In contrast to the two previous studies1,2 assessing corneal aberration in former extreme preterm children with postnatal occurrence of ROP our data reflects rather the effects of low birth in a population-based setting in adults. A further limitation is that no ocular wavefront analysis was conducted. Therefore, our study cannot report on optical quality of the human eye and we cannot state whether the increase of higher-order aberrations leads to decreased image quality on the retina in adulthood. Furthermore, corneal HOA in middle-aged adults could be affected by other factors than low BW, and we cannot state on causal relationships. Another limitation is that the presence of corneal ectasia and corneal dystrophy as well as other ophthalmic diseases apart from surgery and injury was not surveyed and may have affected our results. A further limitation is that contact lens wearers were eligible for the analysis. Participants wearing rigid gas permeable contact lenses may be expected to have some degree of corneal warpage that affects the aberrations measured. Thus we performed a sensitivity analysis and excluded persons with any contact lens use which resulted in comparable results. Limitations with respect to corneal imaging are the scanning protocol with 25 Scheimpflug images used in the present study. A previous validation study43 showed slightly poorer repeatability than, and reduced agreement with, Placido-disk videokeratoscopy particularly for certain aberrations with higher azimuthal frequenties using a scanning protocol of 50 Scheimpflug images. This has to be considered when interpreting our results.

Strengths of our study are the population-based study design, the large sample size, strict standardized examinations of all participants, and all investigators were masked to participants’ birth weight.