|Year : 2021 | Volume
| Issue : 1 | Page : 5-10
A comparative study on the accuracy of SRK/T, holladay II, and barrett universal II formulas in intraocular lens power calculation of axial myopes undergoing cataract surgery
Pradeep Sune, Mona Sune, Ritica Mukherji
Department of Ophthalmology, Jawaharlal Nehru Medical College, Datta Meghe Institute of Medical Sciences (Deemed to be University), Wardha, Maharashtra, India
|Date of Submission||21-Oct-2020|
|Date of Decision||10-Jan-2021|
|Date of Acceptance||25-Jan-2021|
|Date of Web Publication||29-Jul-2021|
Dr. Ritica Mukherji
T-28 Shalinata P.G Girls Hostel, Datta Meghe Institute of Medical Sciences, Sawangi (M), Wardha - 442 001, Maharashtra
Source of Support: None, Conflict of Interest: None
Context: Calculation of intraocular lens (IOL) power for extremes of axial lengths (ALs), using classical formulae like Sanders-Retzlaff-Kraff/Theoretical (SRK/T), is unsatisfactory. With the advent of fourth-generation formulae, surgeons are now aiming for better refractive outcomes for such patients postcataract extraction. Aims: The aim of this study was to assess and compare the accuracy of SRK/T, Holladay II, and Barrett Universal II formulae for IOL power calculation in high myopes with cataract by computing the mean absolute error (MAE). Settings and Design: This was a hospital-based prospective comparative study conducted over a period of 1 year. Forty patients with AL ≥26 mm undergoing cataract extraction were included after taking inclusion and exclusion criteria into consideration. Subjects and Methods: After enrolling patients and obtaining informed consent, all patients underwent a comprehensive ophthalmic examination. The power of IOL to be implanted was calculated by SRK/T, Holladay II, and Barrett Universal II with the goal of achieving refraction within ± 1 D postoperatively. Phacoemulsification was performed for all patients and hydrophilic foldable IOLs were implanted. Patients were followed up for 2 months at the end of which MAE was calculated for all three formulae. Statistical Analysis: Statistical analysis was done by descriptive and inferential statistics using Kruskal–Wallis Chi-square test and Mann–Whitney U-test. Software used in the analysis was SPSS 24 version and P < 0.5 was considered as the level of significance. Results: Barrett Universal II formula had the lowest MAE with a mean of 0.06 ± 0.20 and a median absolute error of 0.02. This was followed by Holladay II and SRK/T. While Barrett Universal II outperformed for all powers of IOL, SRK/T had lower MAE for plus power IOLs and Holladay II was better for negative-power IOLs. Conclusions: Barrett Universal II formula is ideal for IOL power calculation in cataract patients with AL >26 mm.
Keywords: Barrett Universal II, high myopia, Holladay II, mean absolute error, SRK/T
|How to cite this article:|
Sune P, Sune M, Mukherji R. A comparative study on the accuracy of SRK/T, holladay II, and barrett universal II formulas in intraocular lens power calculation of axial myopes undergoing cataract surgery. J Datta Meghe Inst Med Sci Univ 2021;16:5-10
|How to cite this URL:|
Sune P, Sune M, Mukherji R. A comparative study on the accuracy of SRK/T, holladay II, and barrett universal II formulas in intraocular lens power calculation of axial myopes undergoing cataract surgery. J Datta Meghe Inst Med Sci Univ [serial online] 2021 [cited 2023 Sep 25];16:5-10. Available from: https://journals.lww.com/dmms/pages/default.aspx/text.asp?2021/16/1/5/322634
| Introduction|| |
High axial myopes with cataract pose multiple challenges for ophthalmologists in terms of preoperative considerations such as improper axial length (AL) calculations due to poor fixation or posterior staphylomas which affect the calculation of intraocular lens (IOL) power, intraoperative complications such as suprachoroidal hemorrhage, retinal tears or detachments, and posterior capsular rents. Postoperatively, surprise refractive outcomes, especially hypermetropia, are troublesome for the patients. While intraoperative deliberations vary with surgeons, postoperative refractive surprises can be countered by implanting appropriate lens power.
The prevalence of high myopia in central India has been found to be 0.53%. Multiple studies have shown conclusively that high myopia or pathological myopia is associated with faster progression to cataract.,,,,, This could be attributed to the longer ALs affecting nutrient delivery to the posterior aspect of the lens.
This study was conducted to evaluate the predictive accuracy of three IOL formulas belonging to the third and fourth generation in a subset of the Indian population of central India as there is very limited literature available on the same for this geographic area.
Third-generation formulas such as SRK/T have been found to give accurate results for medium-sized eyeballs. This generation makes use of constants to find effective lens position (ELP), i.e., the distance between the anterior corneal plane and the IOL plane. Other formulas in this generation are Hoffer Q and Holladay I. Fourth-generation formulas include Holladay II, Barrett Universal formula II, Olsen formula, Haigis, and Hoffer H. They have a separate set of constants and also consider anterior chamber depth (ACD) for better assessment of ELP.
The aim of this study was to gauge the accuracy of SRK/T, Holladay II, and Barrett Universal II formula in calculating IOL power to achieve a postoperative refraction in the range ± 1 D in high myopes (AL > 26 mm) undergoing cataract surgery. The primary objective was to review and compare the mean absolute error (MAE) for all three formulae, while the secondary objective was to compare the same in the three patient groups based on the power of IOL implanted.
| Subjects and Methods|| |
The design of this study was prospective comparative.
The study was conducted at Department of Ophthalmology in a Teaching Tertiary Care Hospital in Wardha, Maharashtra.
The study was conducted over a period of 1 year from April 2019 to April 2020.
The sample size was 50 eyes of 40 patients.
Subjects and selection methods
A total of 40 patients were evaluated. The patients with AL >26 mm were sequentially selected after taking inclusion and exclusion criteria into consideration. Of 40 patients, 10 patients underwent cataract extraction in both eyes which had AL >26 mm. Therefore, a total of 50 eyes were assessed.
- Patients with AL ≥26 mm
- Patients with uncomplicated senile cataract.
- Patients not giving consent
- Patients where AL could not be determined with absolute clarity
- Patients lost to follow-up (up to 2 months postoperatively)
- Patients with preexisting anterior and posterior segment ocular pathology that could affect refraction postoperatively such as corneal dystrophies, keratoconus or keratoglobus, corneal scars, retinal detachment, or macular edema
- Patients with intraoperative complications that could affect refractive status postoperatively like loss of capsular support requiring sulcus or anterior chamber placement of IOL, vitreous loss, etc.
- Patients with postoperative ocular pathologies affecting refraction
- Patients with systemic diseases such as uncontrolled diabetes mellitus and hypertension, who could not be taken for cataract surgery
- Uncooperative patients or patients with psychiatric illnesses.
The study was approved by the Institutional Ethics Committee (IEC) of the University (Ref. DMIMS (DU)/IEC/2019/7975) and was carried out per the Declaration of Helsinki.
After obtaining informed consent, all subjects underwent a thorough ophthalmic evaluation including best-corrected visual acuity, intraocular pressure measurement, slit-lamp biomicroscop, y and fundus examination.
All patients underwent keratometry using Topcon KR-8000 auto keratometer. ACD, lens thickness (LT), and AL measurement was done using A-scan applanation ultrasonography (PacScan Plus by Sonomed, Model 300A+).
On fundus examination, three subjects were diagnosed to have an eccentric posterior staphyloma. For these patients as well as patients with highly dense cataract where direct visualization of fundus was not possible, AL measurement was done using both A scan and B scan. The visualization of the fovea allowed retinal peak of A scan to coincide with it and give a true measurement of AL.
Calculation of IOL power was done using SRK/T formula software available on the A-scan machine as well as back-calculation using Holladay II formula (https://www.hic-soap.com/calc/) and Barrett Universal formula II (https://calc.apacrs.org/barrett_universal2105/) software which was freely available online. AL adjustment was not done for the latter formulae. White-to-white distance was left unspecified. The goal of the surgeon was to achieve a postoperative refraction within the range of ± 1 D spherical equivalent (SE) which was the primary deciding factor in selecting of IOL power to be implanted. If there was a difference of more than 0.5 D in between the formulas, the IOL power given by SRK/T formula was implanted Furthermore, the surgeon preferred a myopic predictive outcome over a hypermetropic outcome whenever possible.
All patients underwent an uncomplicated 3–3.5 mm sutureless phacoemulsification within the bag IOL implantation and were given standard postoperative care. Patients were followed up for a period of 2 months and refraction was done using the same Topcon KR-8000 auto keratometer at the end of 2 months.
The MAE was calculated and compared for all three formulae to determine accuracy., Furthermore, patients were divided into three groups based on the power of IOL implanted, i.e., plus, minus, or zero power. The MAE of all three formulae was compared for all three groups.
Statistical analysis used
Statistical analysis was done by descriptive and inferential statistics using Kruskal–Wallis Chi-square test and Mann–Whitney U-test and software used in the analysis was SPSS 24.0 version (Chicago, Illinois, USA) and Graph Pad Prism 7.0 version (GraphPad Software, San Diego, CA, USA) and P < 0.05 is considered as the level of significance.
The Institutional Ethics Committee of DMIMSDU has approved the Research work proposed to be carried out at Jawaharlal Nehru Medical College, Sawangi(M), Wardha. Date: 5th April 2019 with Reference no DMIMS (DU)/IEC/2019/7975.
| Results|| |
The maximum age of the patients in the study was in the range of 60–69 years, i.e., 45%. Only one patient was seen in the age group of 10–19 years. Both eyes of this patient were included in the study accounting for 3% of the sample size. The mean age was 54.35 ± 13.07 years and ages ranged from 10 to 76 years [Table 1] and [Figure 1].
About 32.5% of the patients were male and 67.5% were female with a male-to-female ratio of 0.48 [Table 2] and [Figure 2].
The mean corneal power (K) of eyes was 45.16 ± 2.91 and values ranged from 40.00 to 53.00. The mean ACD for all eyes was 4.00 ± 0.42 mm with a minimum ACD of 3.20 mm and a maximum ACD of 4.97 mm. The average AL for 50 eyes was 28.39 ± 2.08 with a range from 26.06 to 34.67. The mean preoperative spherical dioptric power recorded for all patients was − 10.14 DS and values ranged from − 4.00 DS to −20.00 DS [Table 3] and [Figure 3].
The MAE for each of the three formulae was calculated as the difference between the preoperative predictive and postoperative actual SE. SE was computed by adding half the cylindrical power to the spherical power of the patient. Kruskal–Wallis Chi-square test was employed to compare the above. Barrett Universal II formula showed the lowest MAE as compared to the other two formulae with a mean of 0.06 ± 0.20 and a median absolute error of 0.02. The values of absolute error for this formula ranged from −0.51 to 0.41.
This was followed by Holladay II with an average of 0.14 ± 0.03. The median absolute error was found to be 0.21 and the difference between the true and predicted preoperative refraction ranged from −0.90 to 0.82.
SRK/T disclosed an MAE of 0.28 ± 0.24 with a median value of 0.32 and a range between −0.92 and 0.76 [Table 4] and [Figure 4].
The between-group difference for the three formula was statistically significant with P = 0.0001. Thus, to further clarify the relationship between the three formulas employed, the nonparametric Mann–Whitney U-test was used to compare the formulae with each other independently.
Barrett Universal II when compared with Holladay II and SRK/T individually also showed statistical significance with P = 0.0001 in the Barrett Universal II–SRK/T and in the Barrett Universal II–Holladay II group [Table 5]. Furthermore, there was a statistically significant between-group difference in the SRK/T-Holladay II group with P = 0.049 [Table 5].
Patients were further grouped according to the power of IOL which was ultimately implanted. All three formulae and their MAE were analyzed according to these groups.
Of 50 eyes, 36 eyes were implanted with a plus power IOL, 11 eyes with minus power, and 3 eyes with zero power IOLs. The IOL powers implanted ranged from −14 D to +16.5 D.
SRK/T had a mean of 0.24 ± 0.29 for values of MAE in patients implanted with plus power IOLs. The MAE ranged from −0.92 to 0.73 for these patients. Similarly, the mean for minus power IOLs and zero power IOLs was 0.31 ± 0.27 and 0.62 ± 0.12, respectively, whereas the ranges lay between −0.24 to 0.72 and 0.52 to 0.76, respectively. The difference between the three groups was statistically significant with P = 0.038.
Holladay II had the lowest mean for minus power IOLs, i.e., 0.05 ± 0.49 followed by plus power IOLs with a mean of 0.13 ± 0.32 and zero power IOLs with an average of 0.52 ± 0.02. The difference for all three groups was statistically significant with P = 0.044. This formula generated MAE in the range between −0.77 to 0.68 and −0.90 to 0.82 for plus and minus power IOLs, respectively. The range for zero power IOLs was 0.50–0.55.
Barrett Universal II formula yielded the lowest average among the three formulae for all powers of IOL implanted with a mean of −0.09 ± 0.19, −0.02 ± 0.17, and 0.29 ± 0.10 for plus, minus, and Plano IOLs, respectively. The range of MAE was found to be −0.51 to 0.20 for plus power, −0.25–0.38 for minus power, and 0.22 to 0.41 for Plano IOLs [Table 6] and [Figure 5].
| Discussion|| |
With the advent of newer IOL power calculation formulae, it has become comparatively easier to the postoperative refraction of the patient and modifies the type and power of IOL to be inserted based on the desired refractive outcome. This is especially important in the current scenario where patients undergoing cataract surgery expect and demand perfect vision post surgery. The process becomes more challenging in patients with extremes of ALs. Multiple studies in the west have compared the different generation of formulas to understand how these formulae behave with increasing and decreasing ALs.
Abulafia et al. conducted a retrospective case series analysis on 106 eyes with AL >26 mm and concluded that in patients implanted with IOL of power ≥ 6 D, third-generation formulae SRK/T, Hoffer Q, Barrett Universal II, Olsen II, Haigis, and Holladay II were performed at par with each other with a prediction error of ±1 D in 93% eyes, whereas in eyes implanted with IOL of power <6 D, performance of Barrett Universal II, Holladay I, and Haigis formula using AL-adjusted method was superior. Wang et al. in 2018 demonstrated through their meta-analysis involving 4047 eyes that Barrett Universal II gave the maximum number of patients with prediction error within ±0.5D in patients with ALs longer than 26 mm. Stopyra W and associates concluded from their study, involving 81 eyes of 70 patients with axial lengths >25mm, that Barrett Universal II gave the lowest average and median of absolute error of prediction as compared to all three third generation formulae, Holladay II and Haigis formula. Similarly, Zhang et al. found Barrett Universal II to have the least predictive error compared to third-generation formulae and Haigis formula.
These are in contrast to the studies conducted by Bang et al. and Terzi et al. who stated that Haigis formula is most accurate in calculating IOL power in long eyes as compared to Holladay II and SRK/T as well as the studies conducted by Tang et al. and Wan et al. who found no significant between-group difference in calculating IOL power for high myopes by Hill RBF, Holladay II, and Barrett Universal II formulae.
In the present study as well, Barrett Universal II formula had a lower average MAE as compared to SRK/T and Holladay II formula. When compared with each formula independently, the difference in MAE was still statistically significant and found to be lower than the other two formulae and for all powers of IOLs implanted. This could be attributed to the use of AC depth and LT in the formula which allows a more precise calculation of power. This resounds well with other studies which also prove Barrett Universal formula to be superior to others in IOL power calculation of high myopes with cataract.,,,
The performance of Holladay II was found to be better for minus power IOLs in our study and almost comparable to the performance of Barrett Universal II. It performed better than SRK/T in terms of MAE as well as had lower predictive errors for all powers of IOLs implanted. This correlates well with the findings of Melles et al. who also found that the Holladay II formula was better than SRK/T for predictive accuracy in higher myopes with cataract.
Zaldivar et al. found that 92% of patients had a postoperative refraction in the range of ±1 D in case of plus power IOLs when SRK/T formula was used and 41% of patients had the same in case of minus power IOLs when Holladay II was used, which was also seen in this study. Ghanem and El-Sayed also concluded that Holladay II had a better performance than SRK/T for longer ALs.
Of the three formulae evaluated, SRK/T had the highest MAE for all powers of IOL and among the other two formulae. Anand et al. had put forward that SRK/T was the most accurate formula for calculation of power in patients with ALs >24.5; however, the study had not taken into account fourth-generation formulae like Holladay II or Barrett Universal formula II. Doshi et al. also found predictability of SRK/T, Hoffer Q, Holladay I, and Haigis to be comparable for ALs >24 mm.
One of the main limitations of our study was that the MAE for the formulae was not correlated to the ALs. It can be argued that most patients receiving minus power IOLs are found to have longer ALs as compared to those who receive plus power IOLs. Zhou et al. found that SRK/T and Barrett Universal II formula had better predictive outcomes for plus-powered IOLs than minus power IOLs, but Barrett Universal II still had an advantage over other formulas for such eyes, which is also reflected in our study. Another limitation was the low number of patients implanted with negative or zero power IOLs, but this reflects the actual patient demographic encountered with high myopes even though the surgeon made a conscious effort to aim for myopia rather than hyperopia whenever possible.
We recommend further large-scale studies comparing different constants, for example, ULIB constants and with different AL adjustments, such as those seen in the Holladay II software, in very long eyes for more accurate results. Fifth-generation formulae like Hill-RBF formula should also be evaluated as the literature for the same is very limited in the Asian-Indian population.
| Conclusions|| |
Barrett Universal II formula had the highest predictive accuracy in patients with high myopia and cataract. Holladay II has better outcomes for minus power IOLs and may be used for very long ALs.,, [30,,
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Gologorsky D, Flynn HW Jr. Cataract surgery in the setting of severe pathologic myopia with high axial length: use of pars plana lensectomy and vitrectomy. Clin Ophthalmol 2016;10:989-92.
Nangia V, Jonas JB, Sinha A, Matin A, Kulkarni M. Refractive error in central India: The central India eye and medical study. Ophthalmology 2010;117:693-9.
Duke Elder S. System of ophthalmology. Vol. 11. London: Kimpton; 1970. p. 225.
Von Kluxen G. Klinische und experimentelle untersuchungen an alterskatarakten. Fortschr Med 1985;103:243-6.
Perkins ES. Cataract: Refractive error, diabetes, and morphology. Br J Ophthalmol 1984;68:293-7.
Racz P. Koszoras L, Ordogh M. Investigations on bilateral presenile cataracts. Lens Res 1984;2:13-21.
Weale R. A note on a possible relation between refraction and a disposition for senile nuclear cataract. Br J Ophthalmol 1980;64:311-4.
Younan C, Mitchell P, Cumming RG, Rochtchina E, Wang JJ. Myopia and incident cataract and cataract surgery: The blue mountains eye study. Invest Ophthalmol Vis Sci 2002;43:3625-32.
Anand N, Khurana A, Bhardwaj K, Deswal J, Verma S. A comparative study of intraocular lens power calculation formulae (SRK/T, Hoffer Q, SRK II AND Haigis) in eyes of patients with various axial lengths, EJPMR, 2018,5(9), 260-5.
Amro M, Chanbour W, Arej N, Jarade E. Third- and fourth-generation formulas for intraocular lens power calculation before and after phakic intraocular lens insertion in high myopia. J Cataract Refract Surg 2018;44:1321-5.
Aristodemou P, Knox Cartwright NE, Sparrow JM, Johnston RL. Statistical analysis for studies of intraocular lens formula accuracy. Am J Ophthalmol 2015;160:1085-6.
Hoffer KJ, Aramberri J, Haigis W, Olsen T, Savini G, Shammas HJ, et al
. Protocols for studies of intraocular lens formula accuracy. Am J Ophthalmol 2015;160:403-50.
Abulafia A, Barrett GD, Rotenberg M, Kleinmann G, Levy A, Reitblat O, et al
. Intraocular lens power calculation for eyes with an axial length greater than 26.0 mm: Comparison of formulas and methods. J Cataract Refract Surg 2015;41:548-56.
Wang Q, Jiang W, Lin T, Zhu Y, Chen C, Lin H, et al
. Accuracy of intraocular lens power calculation formulas in long eyes: Asystemic review and metaanalysis. Clin Exp Ophthalmol 2018;46:738-49.
Stopyra W. The accuracy of intraocular lens power calculation formulas for eyes of axial length exceeding 25.0 mm. Austin J Surg 2019;6:1227.
Zhang Y, Liang XY, Liu S, Lee JWY, Bhaskar S, Lam DSC. Accuracy of Intraocular Lens Power Calculation Formulas for Highly Myopic Eyes. Macky TA, editor. Journal of Ophthalmology. 2016; 2016:1917268.
Bang S, Edell E, Yu Q, Pratzer K, Stark W. Accuracy of intraocular lens calculations using the IOLMaster in eyes with long axial length and a comparison of various formulas. Ophthalmology 2011;118:503-6.
Terzi E, Wang L, Kohnen T. Accuracy of modern intraocular lens power calculation formulas in refractive lens exchange for high myopia and high hyperopia. J Cataract Refract Surg 2009;35:1181-9.
Tang KS, Tran EM, Chen AJ, Rivera DR, Rivera JJ, Greenberg PB. Accuracy of biometric formulae for intraocular lens power calculation in a teaching hospital. Int J Ophthalmol 2020;13:61-5.
Wan KH, Lam TC, Yu MC, Chan TC. Accuracy and precision of intraocular lens calculations using the new hill-RBF version 2.0 in eyes with high axial myopia. Am J Ophthalmol 2019;205:66-73.
Barrett GD. An improved universal theoretical formula for intraocular lens power prediction. J Cataract Refract Surg 1993;19:713-20.
Zhou D, Sun Z, Deng G. Accuracy of the refractive prediction determined by intraocular lens power calculation formulas in high myopia. Indian J Ophthalmol 2019;67:484-9.
] [Full text]
Kuthirummal N, Vanathi M, Mukhija R, Gupta N, Meel R, Saxena R, et al
. Evaluation of Barrett universal II formula for intraocular lens power calculation in Asian Indian population. Indian J Ophthalmol 2020;68:59-64.
] [Full text]
Melles RB, Holladay JT, Chang WJ. Accuracy of intraocular lens calculation formulas. Ophthalmology 2018;125:169-78.
Zaldivar R, Shultz MC, Davidorf JM, Holladay JT. Intraocular lens power calculations in patients with extreme myopia. J Cataract Refract Surg 2000;26:668-74.
Ghanem AA, El-Sayed HM. Accuracy of intraocular lens power calculation in high myopia. Oman J Ophthalmol 2010;3:126-30.
] [Full text]
Doshi D, Limdi P, Parekh N, Gohil N. A comparative study to assess the predictability of different iol power calculation formulas in eyes of short and long axial length. J Clin Diagn Res 2017;11:NC01-4. 1. Bajpayee N., P.K. Tidake, and P.M. Palsodkar. “Comparative Study of Visual Acuity and Astigmatism Determination between Small Incision Cataract Surgery and Phacoemulsification by Corneal Topography.” International Journal of Current Research and Review 2020;12:65-8. Available from: https://doi.org/10.31782/IJCRR.2020.SP82
. [Last accessed on 2020 July 21].
Gu A, Daigavane S, Palsodkar PM. Effect of Heparinised Irrigating Solution during Cataract Surgery on Postoperative Inflammation. Int J Pharm Res 2019;11:1800-3. Available from: https://doi.org/10.31838/ijpr/2019.11.02.200
. [Last accessed on 2020 July 21].
Muley S, Saoji C, Daigavane S. The Effect of Intracameral Adrenaline Bolus Injection on Pupil Size, Pulse Rate and Blood Pressure during Small Incision Cataract Surgery. Int J Cur Res Rev 2020;12:114-8. Available from: https://doi.org/10.31782/IJCRR.2020.114118
. [Last accessed on 2020 July 21].
Muley S, Saoji C, Daigavane S, Patil R, Ghogare AS. Anxiety in Cataract Patients, Benefits of Counseling on Patient Anxiety in Routine Cataract Surgery. Int J Res Pharm Sci 2020;11:2620-5. Available from: https://doi.org/10.26452/ijrps.v11i2.2272
. [Last accessed on 2020 July 21].
Nimbulkar G, Wagh V, Gaidhane A, Chhabra KG, Deolia S, Ingole A. A Study to Determine the Role of Different Types of Intraocular Lenses in the Management of Traumatic Cataract in a Tertiary Healthcare Centre. Eur J Molecular Clin Med 2020;7:2035-9.
Shekhar S, Tidake PK. Comparative Study of Visual Outcome, Astigmatism, and Complications between Conventional Extracapsular Cataract Extraction and Manual Small Incision Cataract Surgery. J Datta Meghe Inst Med Sci Univ 2020;15:215-22. Available from: https://doi.org/10.4103/jdmimsu.jdmimsu_201_20
. [Last accessed on 2020 July 21].
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6]