Intraocular lens (IOL) power calculation is a crucial step in achieving the desired target refractive outcome, which is a major aim of modern day cataract surgery. Numerous devices and formulas are currently available, allowing accurate determination of the IOL power needed to reach the target refraction [1–3,4▪▪,5▪,6–9,10▪,11–16,17▪,18]. In order to accomplish target refraction, axial length, anterior chamber depth (ACD), and corneal radii (K1 and K2) need to be accurately measured. Moreover, proper choices of IOL power calculation formulas are important, as are the use of accurate IOL constants, depending on the type of IOL and postoperative IOL location. Over the past decade, significant developments have been made, which have led to improvements in the predictability of the refractive outcomes. These include stable in-the-bag IOL placement that ensures stability and more predictable IOL positioning, as well as modifications in IOL power calculation formulas [1,2]. In addition, recent developments in the biomedical field have led the availability of novel devices, such as the laser partial coherence interferometry (PCI) and the low-coherence optical reflectometry (LCOR) [15,19]. To date, A-mode ultrasound biometry had been considered the gold standard for axial length and ACD measurement. The PCI-based IOLMaster (Carl Zeiss Meditec AG, Jena, Germany) was introduced in 1999. More recently, a new biometry device, the Lenstar LS 900 (Haag Streit AG, Bern, Switzerland) using LCOR technology was introduced in 2008. Given the heightened patient expectations, it is of utmost importance to accurately predict the correct IOL power. The recent technological developments have stimulated continuous modifications in biometry. This article reviews recent studies and advances in the field of clinical biometry.
CONTACT ULTRASOUND OCULAR BIOMETRY
A-mode contact ultrasound ocular biometry has been considered the gold standard for decades. A special crystal embedded in a probe oscillates to generate a high-frequency sound wave that penetrates the eye. This results in a one-dimensional time-amplitude representation of echoes received along the beam path. The distance between the echo spikes recorded on the oscilloscope screen provides an indirect measurement of tissue such as globe length or lens thickness. The height of the spike is indicative of the strength of the tissue sending back the echo. There are two types of A-mode ultrasound biometry available, including contact applanation biometry and immersion biometry. Contact type biometry requires a probe placed on the cornea and is prone to errors due to corneal indentation and off-axis measurements. It also carries risk of transmitting infections. Immersion-type biometry requires placing a saline-filled scleral shell between the probe and the eye. As no pressure is applied on the eye, corneal indentation is prevented.
NONCONTACT OPTICAL BIOMETRY
Optical biometry for accurate assessment of the axial length is increasingly becoming popular, as it is rapid, easy to use, and a contact-free method. The PCI-based IOLMaster uses a 780 nm laser diode infrared light to measure axial length (Table 1) . The ACD is measured through a lateral slit illumination with this device, and the anterior corneal curvature is calculated at six reference points in a hexagonal pattern at approximately the 2.3 mm optical zone. The new Lenstar LS 900 is LCOR-based and uses an 820-nm superluminescent diode (Table 1) . In addition to axial length, it measures central corneal thickness (CCT), as well as lens thickness. ACD measurements differ between the IOLMaster and the Lenstar, as Lenstar measures ACD from the corneal endothelium to the anterior lens surface, whereas IOLMaster measures ACD from corneal epithelium to the anterior lens surface. The Lenstar also measures crystalline lens thickness and retinal thickness, as well as the size and centricity of the pupil . K readings are calculated by analyzing the anterior corneal curvature at 32 reference points orientated in two circles at approximately the 2.30 and 1.65 mm optical zones.
These IOLMaster and Lenstar LS 900 are in good agreement in terms of mean axial length, ACD, and K readings (Table 2) [9,19–25]. The mean difference in axial length measurements was only 0.01 mm ± 0.05 (SD) between these two devices (P = 0.12). The LCOR-based device measures more parameters than the PCI-based device, including CCT, retinal thickness and pupil diameter. Whereas this is an advantage, measurements with the LCOR-based device take twice as long as those with PCI-based device . Although these optical biometry devices are easy to use, their main disadvantage, in our experience, is the failure to provide axial length measurements in dense subcapsular cataracts. As they use a laser beam, rather than an ultrasound wave, dense cataracts do not allow the laser beam to reach the retina and reflect accordingly. This issue has previously been reported in several other studies [19,25]. Thus, we recommend having an A-scan ultrasound biometry device, in order to overcome this obstacle.
REFRACTIVE POWER MEASUREMENT OF THE CORNEA
Measuring corneal power is always puzzling, as neither manual nor automated keratometers can directly measure the ‘true’ corneal power. Instead, the cornea is assumed to be a spherocylinder with a fixed anterior-to-posterior corneal curvature. A very fundamental problem in the design of manual and automated keratometers is that they do not provide sufficient information to determine corneal shape accurately. A well lit target is placed in front of the cornea, which acts as a convex mirror and produces a virtual image of the target. The corneal radius of curvature may then be predictable from this, provided several statements are valid (the cornea is presumed to be spherical; paraxial optics is assumed; the power of the back corneal surface is estimated). Nonetheless, the obtained results are very diminutive. Whereas in most healthy eyes corneal power is relatively easily calculated [26▪▪], eyes with prior refractive surgery pose a particular challenge. With the change in anterior and posterior corneal curvature in these eyes, the corneal refractive index (n = 1.3375) is no longer accurate. Thus, the corneal power is underestimated in these eyes due to corneal flattening, which in turn leads to overestimation of corneal power and a hyperopic refraction after cataract surgery [26▪▪,27].
Computerized videokeratography may be superior to manual keratometers in assessing corneal power in postrefractive surgical eyes. Holladay et al. evaluated the accuracy of central corneal power measurements by Scheimpflug imaging (Pentacam) for eyes that had undergone refractive surgery. They used historical method to compute the theoretical postoperative keratometry (K)-reading, which was then compared with the measured equivalent K-reading (EKR) from the Pentacam. The mean prediction error for the pilot group was −0.06 ± 0.56 diopters (D). Using the 4.5-mm zone determined in the pilot group, the EKR value for the test group of 41 radial keratotomy eyes had a mean prediction error of −0.04 ± 0.94 D (range: −1.84 to ± 2.27 D). They concluded that Scheimpflug imaging with the Pentacam provides an alternative method of measuring the central corneal power in eyes that previously received corneal refractive surgery. In another study by Tang et al., accuracy of Pentacam EKR readings was found to be inaccurate in virgin corneas as well as in those with a history of laser in-situ keratomileusis (LASIK), photorefractive keratectomy (PRK), or radial keratotomy. The Pentacam power measurements were consistently steeper than the true corneal power. Kim et al. used true net corneal power of the Pentacam system to provide keratometry readings in eyes, which had previously undergone refractive surgery. The mean deviation from the desired postoperative cataract refractive outcome was 0.47 ± 0.56 D. Jin et al. calculated corneal power by using Gaussian optics formula in postrefractive surgery eyes. Using this calculated K for IOL power calculation in postrefractive cases yielded mean absolute prediction errors of 0.58 ± 0.52 D (Haigis), 0.59 ± 0.49 D (double-K Hoffer Q), and 0.58 ± 0.47 D (double-K SRK/T). Arce et al. applied the method developed by Sonego-Krone et al. using the Orbscan II in eyes with previous myopic and hyperopic refractive surgery. The corneal power in this study represents the average of all points obtained from the Orbscan II total mean-maps within the central 2-mm diameter as measured directly from corneas with previous refractive surgery. The overall difference between the calculated and achieved refraction (0.12 ± 0.93 D, P = 0.27) was 1.00 D in 77% of eyes and 2.00 D in 96% of eyes. They concluded that keratometer readings could be performed with reasonable accuracy using the Orbscan II central 2-mm total mean power in eyes with previous corneal refractive surgery. However, this method yielded better outcomes in eyes with previous radial keratotomy, myopic LASIK, and myopic PRK, as compared with eyes with hyperopic LASIK or radial keratotomy with LASIK. In addition, the average power from the central 4-mm zone of the total-optical map also reflected accurately the refractive change after myopic LASIK .
Reduced accuracy of IOL calculations after corneal refractive surgery is a clinical problem of growing importance. There are several methods in the literature to evaluate the corneal power after refractive surgery, including the clinical history method (vertex corrected to the corneal plane), the contact lens over-refraction method, and the Aramberri double-K method, the Latkany Flat-K. Although these methods offer better accuracy in postrefractive surgery eyes, preoperative and postoperative K values and/or refraction are still required before cataract surgery, which is time-consuming to perform. To save time, Wang et al.[33▪▪] developed an Internet-based IOL power calculator for eyes with previous LASIK, PRK, or radial keratotomy. Methods using pre-LASIK/PRK keratometry (K) and surgically induced change in refraction, methods using surgically induced change in refraction, and methods using no previous data were evaluated. They found that methods using only surgically induced changes in refraction resulted in superior outcomes as compared to methods using pre-LASIK/PRK K values and surgically induced change in refraction. In a recent study by McCarthy et al.[34▪▪] methods of IOL power calculation after myopic laser refractive surgery were compared in a large, multisurgeon study. The top five corneal power adjustment techniques and formula combinations were the Masket with the Hoffer Q formula, the Shammas.cd with the Shammas-PL formula, the Haigis-L, the Clinical History Method with the Hoffer Q, and the Latkany Flat-K with the SRK/T. They concluded that, by using these methods, 70–85% of eyes could achieve visual outcomes within 1.0 D of target refraction.
Taken together, it has been widely suggested that it is logical to use several different methods in determining corneal power rather than trusting simply on any one method alone.
ANTERIOR CHAMBER DEPTH MEASUREMENT
Most of the modern IOL power formulas depend on ACD measurements in order to increase the accuracy of the IOL power prediction curve. Hence, accurate measurements are crucial to lessen the possibility of undesirable refractive outcomes. All modern biometry devices, as well as slit-scanning videokeratography (Orbscan), Scheimpflug imaging (Pentacam), and anterior segment optical coherence tomography (AS-OCT) are capable of measuring ACD.
There are several studies, which compared ACD measurements between various biometry devices (Table 2) [20,22,35–38]. Salouti et al. compared ACD readings obtained by Lenstar LS 900, IOLMaster, and A-mode biometry. The ACD measurements obtained by IOLMaster were slightly smaller than those obtained with other devices, demonstrating 3.14 ± 0.40 mm (range: 2.30–4.27 mm) with A-mode biometry; 3.07 ± 0.42 mm (range: 2.10–4.16 mm) with IOLMaster; 3.17 ± 0.42 mm (range: 2.19–4.51 mm) with Lenstar. However, these differences were not statistically significant (P = 0.09). The mean difference was 0.07 ± 0.19 mm between A-mode biometry and IOLMaster; −0.03 ± 0.24 mm between A-mode biometry and Lenstar, and −0.10 6 0.26 mm between IOLMaster and Lenstar. It was found that Lenstar LS 900 consistently gave slightly higher ACD readings, although not clinically significant, compared with those of the IOLMaster. This finding is similar to other studies in which a similar difference was found [9,19,24]. Chen et al. compared ACD measurement performed by IOLMaster with the present gold standard A-mode biometry. They found that mean ACD reading was 3.13 ± 0.58 mm (range: 2.26–5.14 mm) with IOLMaster and 3.12 ± 0.49 mm (range: 2.27–4.10 mm) with A-mode biometry. There was a very small difference of 0.00 ± 0.05 mm between these two devices. One of the limitations of that study is that although they included Lenstar LS 900 in their study, they did not include mean ACD readings obtained by this device in their analyses. Salouti et al. reported a mean difference of 0.32 and 0.30 mm between Orbscan-Galilei (dual Scheimpflug system; Zimmer Ophthalmics, Port, Switzerland) and Orbscan-Pentacam, respectively. However, this difference was only 0.02 mm between Pentacam and Galilei. These data indicate that Orbscan gives consistently higher measurements for anterior chamber depth compared with Galilei and Pentacam. Therefore, they are not interchangeable in every clinical situation. It is important to note that the ACD measuring module of the IOLMaster should not be used to determine the ACD in a pseudophakic eye, as the evaluation software is only designed for phakic eyes. The evaluation algorithms expect scattered light from the crystalline lens, whereas IOLs essentially produce strong reflections, which will be interpreted erroneously. Pseudophakic ACDs will thus give results, which are faulty and unreliable, as is described in the IOLMaster user manual .
AXIAL LENGTH MEASUREMENT
Differences in axial length measurements have a substantial influence on the final calculated IOL power. There are numerous studies comparing IOLMaster and Lenstar LS 900 with A-mode biometry (Table 2) [9,18,20–24,39]. In a recent study by Jasvinder et al., a strong inter-method agreement (only 0.01 mm axial length difference) was found between IOLMaster and Lenstar in phakic eyes. There was a mean difference of 0.04 mm between Lenstar-immersion biometry and 0.188 mm between Lenstar-A-mode contact biometry. A similar difference was found in mean axial length readings between devices in a recent study by Salouti et al. with good agreement. Montés-Micó et al. reported that measurements between IOLMaster, Lenstar, and immersion biometry were highly correlated for axial length (R = 0.99) in cataract patients. There was no statistically significant difference between devices in terms of mean axial length values.
INTRAOCULAR LENS POWER CALCULATION
Though there are significant improvements in biometry devices in terms of technological developments, there is still an ongoing debate about which IOL power calculation formula best predicts actual postoperative refraction. There is not a single formula, which is suitable for all eyes. Therefore, it is important to know the strengths and weaknesses of the modern IOL power calculation formulas widely used in ophthalmic practice, in order to choose the most appropriate and accurate one that fits to a particular patient. The latest third-generation formulas are Hoffer Q, Holladay, and SRK/T. They all use thin-lens formulas, which reduce IOLs to thin lenses of infinite thickness with only one effective lens plane (ELP). What differentiates these formulas is the method by which the ELP is predicted. In a recent study, Aristodemou et al.[4▪▪] calculated hypothetical prediction errors on prospectively collected data using optimized Hoffer Q, Holladay 1, and SRK/T formulas. The Hoffer Q performed best for axial lengths from 20.00 to 20.99 mm, the Hoffer Q and Holladay 1 for axial lengths from 21.00 to 21.49 mm, and the SRK/T for axial lengths of 27.00 mm or longer. Jin et al. compared the accuracy of the thin-lens and ray-tracing formulas in IOL power calculations in normal and postrefractive surgery eyes. They concluded that thin-lens formulas were as accurate as the ray-tracing method in IOL power calculations in these eyes. However, their mean absolute prediction error (MAE) values were considerably higher than those found in recent studies using optical biometry . Olsen attributed these high MAEs to the incorrect use of the algorithm for the ACD prediction, which today is regarded as the major source of error in IOL power calculation . The only ACD formula that has been based on ACD measurements using the IOLMaster is the Haigis formula. The Haigis-L formula is quite different from these two variable formulas mentioned above in that it uses three different constants and a measured ACD, in order to more accurately determine the ELP. It does not require corneal power values in the calculation formula; hence, errors in measuring K values are avoided. However, the main pitfall is that the constants must be originated by a regression analysis.
The individual steps of the SRK/T formula were examined with reference to a database of biometry and refractive outcomes in 11.189 eyes by Sheard et al.[5▪]. They observed a nonphysiologic behavior in the calculation of corrected axial length and corneal height. They developed the T2 formula using a regression formula for corneal height derived from the development subset and they concluded that any surgeon who uses the SRK/T formula could switch to using the T2 formula and improve refractive outcomes by 10%.
Modern technology has significantly improved our ability to accurately measure ocular biometrical parameters. Hence, today, we are more confident fulfilling patient expectations. However, it is still very important to pay attention to appropriate patient selection, accurate keratometry and biometry, and right IOL power formula selection. Eventually, the highest variable parameter is going to establish the outcome. In order to increase accuracy in ocular biometry practice, one must have sought the following realizations: properly calibrated instrument and an experienced operator, repeating measurements, using optical biometry rather than contact biometry, using last-generation IOL formulas and tailoring the IOL constants accordingly, evaluating refractive outcomes regularly, and following modern up-to-date surgical techniques such as good sizing in capsulorhexis. By following each step carefully in preoperative, intraoperative, and postoperative algorithms, understanding strengths and weaknesses during all these steps, successful outcomes are achievable.
Conflicts of interest
The research was supported by a career development grant K08-EY020575 (P.H.) from the National Institutes of Health, Bethesda, Maryland, USA. P.H. is also the recipient of a Career Development Award from Research to Prevent Blindness.
There are no conflicts of interest.
REFERENCES AND RECOMMENDED READING
Papers of particular interest, published within the annual period of review, have been highlighted as:
- ▪ of special interest
- ▪▪ of outstanding interest
Additional references related to this topic can also be found in the Current World Literature section in this issue (p. 76).
1. Olsen T. Calculation of intraocular lens power: a review. Acta Ophthalmol Scand 2007; 85:472–485.
2. Lee AC, Qazi MA, Pepose JS. Biometry
and intraocular lens power calculation. Curr Opin Ophthalmol 2008; 19:13–17.
3. Jin H, Rabsilber T, Ehmer A, et al. Comparison of ray-tracing method and thin-lens formula in intraocular lens power calculations. J Cataract Refract Surg 2009; 35:650–662.
Aristodemou P, Knox Cartwright NE, Sparrow JM, Johnston RL. Formula choice: Hoffer Q, Holladay 1, or SRK/T and refractive outcomes in 8108 eyes after cataract surgery
by partial coherence interferometry. J Cataract Refract Surg 2011; 37:63–71.
Study that defines which IOL formula gives the most accurate result for in particular axial lengths.
Sheard RM, Smith GT, Cooke DL. Improving the prediction accuracy of the SRK/T formula: the T2 formula. J Cataract Refract Surg 2010; 36:1829–1834.
Study applied a modification to SRK/T formula and improved refractive outcomes by 10%.
6. Olsen T. Intraocular lens power calculation. J Cataract Refract Surg 2009; 35:2176–2177.author reply 2177–2178.
7. Tang M, Li Y, Huang D. An intraocular lens power calculation formula based on optical coherence tomography: a pilot study. J Refract Surg 2010; 26:430–437.
8. Haigis W. IOL power calculations. Ophthalmology 2010; 117:400–401.author reply 401.
9. Holzer MP, Mamusa M, Auffarth GU. Accuracy of a new partial coherence interferometry analyser for biometric measurements. Br J Ophthalmol 2009; 93:807–810.
Hirnschall N, Murphy S, Pimenides D, et al. Assessment of a new averaging algorithm to increase the sensitivity of axial eye length measurement with optical biometry
in eyes with dense cataract. J Cataract Refract Surg 2011; 37:45–49.
Study assessing the latest IOLMaster software to decrease the likelihood of unsuccessful scans in dense cataracts.
11. Wong ACM, Mak ST, Tse RKK. Clinical evaluation of the intraoperative refraction technique for intraocular lens power calculation. Ophthalmology 2010; 117:711–716.
12. Charalampidou S, Cassidy L, Ng E, et al. Effect on refractive outcomes after cataract surgery
of intraocular lens constant personalization using the Haigis formula. J Cataract Refract Surg 2010; 36:1081–1089.
13. Olsen T. Improved accuracy of intraocular lens power calculation with the Zeiss IOLMaster. Acta Ophthalmol Scand 2007; 85:84–87.
14. Latkany RA, Chokshi AR, Speaker MG, et al. Intraocular lens calculations after refractive surgery
. J Cataract Refract Surg 2005; 31:562–570.
15. Drexler W, Findl O, Menapace R, et al. Partial coherence interferometry: a novel approach to biometry
in cataract surgery
. Am J Ophthalmol 1998; 126:524–534.
16. Shammas HJ, Chan S. Precision of biometry
, and refractive measurements with a partial coherence interferometry–keratometry
device. J Cataract Refract Surg 2010; 36:1474–1478.
Rozema JJ, Atchison DA, Tassignon M-J. Statistical eye model for normal eyes. Investig Ophthalmol Visual Sci 2011; 52:4525–4533.
A new statistical eye model that is a good addition to the classic eye models.
18. Tappeiner C, Rohrer K, Frueh BE, et al. Clinical comparison of biometry
using the noncontact optical low coherence reflectometer (Lenstar LS 900) and contact ultrasound biometer (Tomey AL-3000) in cataract eyes. Br J Ophthalmol 2010; 94:666–667.
19. Buckhurst PJ, Wolffsohn JS, Shah S, et al. A new optical low coherence reflectometry device for ocular biometry
in cataract patients. Br J Ophthalmol 2009; 93:949–953.
20. Chen YA, Hirnschall N, Findl O. Evaluation of 2 new optical biometry
devices and comparison with the current gold standard biometer. J Cataract Refract Surg 2011; 37:513–517.
21. Hoffer KJ, Shammas HJ, Savini G. Comparison of 2 laser instruments for measuring axial length. J Cataract Refract Surg 2010; 36:644–648.
22. Salouti R, Nowroozzadeh MH, Zamani M, et al. Comparison of the ultrasonographic method with 2 partial coherence interferometry methods for intraocular lens power calculation. Optometry 2011; 82:140–147.
23. Jasvinder S, Khang TF, Sarinder KKS, et al. Agreement analysis of LENSTAR with other techniques of biometry
. Eye 2011; 25:717–724.
24. Rabsilber TM, Jepsen C, Auffarth GU, Holzer MP. Intraocular lens power calculation: clinical comparison of 2 optical biometry
devices. J Cataract Refract Surg 2010; 36:230–234.
25. Rohrer K, Frueh BE, Walti R, et al. Comparison and evaluation of ocular biometry
using a new noncontact optical low-coherence reflectometer. Ophthalmology 2009; 116:2087–2092.
Savini G, Hoffer KJ, Carbonelli M, Barboni P. Intraocular lens power calculation after myopic excimer laser surgery: clinical comparison of published methods. J Cataract Refract Surg 2010; 36:1455–1465.
Study comparing several methods used to calculate IOL power after myopic excimer laser surgery.
27. Holladay JT, Hill WE, Steinmueller A. Corneal power measurements using Scheimpflug imaging in eyes with prior corneal refractive surgery
. J Refract Surg (Thorofare, N J : 1995) 2009; 25:862–868.
28. Tang Q, Hoffer KJ, Olson MD, Miller KM. Accuracy of Scheimpflug Holladay equivalent keratometry
readings after corneal refractive surgery
. J Cataract Refract Surg 2009; 35:1198–1203.
29. Kim SW, Kim EK, Cho BJ, et al. Use of the pentacam true net corneal power for intraocular lens calculation
in eyes after refractive corneal surgery. J Refract Surg 2009; 25:285–289.
30. Jin H, Holzer MP, Rabsilber T, et al. Intraocular lens power calculation after laser refractive surgery
: corrective algorithm for corneal power estimation. J Cataract Refract Surg 2010; 36:87–96.
31. Arce CG, Soriano ES, Weisenthal RW, et al. Calculation of intraocular lens power using Orbscan II quantitative area topography after corneal refractive surgery
. J Refract Surg 2009; 25:1061–1074.
32. Sonego-Krone S, Lopez-Moreno G, Beaujon-Balbi OV, et al. A direct method to measure the power of the central cornea after myopic laser in situ keratomileusis. Arch Ophthalmol 2004; 122:159–166.
Wang L, Hill WE, Koch DD. Evaluation of intraocular lens power prediction methods using the American Society of Cataract and Refractive Surgeons Post-Keratorefractive Intraocular Lens Power Calculator. J Cataract Refract Surg 2010; 36:1466–1473.
Good comparison of methods used by ASCRS postkeratorefractive IOL power calculator.
McCarthy M, Gavanski GM, Paton KE, Holland SP. Intraocular lens power calculations after myopic laser refractive surgery
: a comparison of methods in 173 eyes. Ophthalmology 2011; 118:940–944.
Good comparison of top IOL calculation approaches after myopic laser refractive surgery.
35. Savini G, Olsen T, Carbonara C, et al. Anterior chamber depth measurement in pseudophakic eyes: a comparison of Pentacam and ultrasound. J Refract Surg 2010; 26:341–347.
36. Gursoy H, Sahin A, Basmak H, et al. Lenstar versus ultrasound for ocular biometry
in a pediatric population. Optometry Vision Sci 2011; 88:912–919.
37. Cleary G, Spalton DJ, Marshall J. Anterior chamber depth measurements in eyes with an accommodating intraocular lens: agreement between partial coherence interferometry and optical coherence tomography. J Cataract Refract Surg 2010; 36:790–798.
38. Salouti R, Nowroozzadeh MH, Zamani M, et al. Comparison of anterior chamber depth measurements using Galilei, HR Pentacam, and Orbscan II. Optometry 2010; 81:35–39.
39. Montés-Micó R, Carones F, Buttacchio A, et al.
Comparison of Immersion ultrasound, partial coherence interferometry, and low coherence reflectometry for ocular biometry
in cataract patients. J Refract Surg 2011; 1–7.
40. Norrby S. Sources of error in intraocular lens power calculation. J Cataract Refract Surg 2008; 34:368–376.