|
| |
 |
|
| ORIGINAL ARTICLE |
|
| Year : 2020 | Volume
: 31
| Issue : 3 | Page : 115-121 |
|
A lower urine white blood cell median can be a predictor of undiscovered urolithiasis in patients with acute urinary tract symptoms
Yuan Chen1, Chun-Hou Liao2, Bing-Juin Chiang1, Hsu-Che Huang3, Yu-Wei Chou1
1 Department of Surgery, Division of Urology, Cardinal Tien Hospital, Taipei, Taiwan
2 Department of Surgery, Division of Urology, Cardinal Tien Hospital; Department of Surgery, Division of Urology, Fu Jen Catholic University Hospital, Taipei, Taiwan
3 Department of Surgery, Division of Urology, Cardinal Tien Hospital; Department of Life Science, College of Science, National Taiwan Normal University, Taipei, Taiwan
| Date of Submission | 06-Feb-2020 |
| Date of Decision | 08-Mar-2020 |
| Date of Acceptance | 07-Apr-2020 |
| Date of Web Publication | 26-Jun-2020 |
Correspondence Address:
Hsu-Che Huang
Department of Surgery, Division of Urology, Cardinal Tien Hospital, No. 362, Zhongzheng Road, Xindian, New Taipei City 23148
Taiwan
Yu-Wei Chou
No. 362, Zhongzheng Road, Xindian, New Taipei City 23148
Taiwan
 Source of Support: None, Conflict of Interest: None  | 1 |
DOI: 10.4103/UROS.UROS_12_20
Purpose: Acute urinary tract infection (UTI) causes symptoms including urgency, frequency, dysuria, and pain in the lower urinary tract. However, these symptoms are not unique to infection. Urolithiasis, injury, neurological disease, and malignancy present with the same symptoms. We aimed to find predictors in urinalysis to detect underlying causes in patients with acute UTI symptoms. Materials and Methods: We recruited 192 patients who presented with acute UTI symptoms from a single institute between September 2012 and September 2018. Each patient received urinalysis, kidney–ureter–bladder X-ray, and kidney plus transabdominal bladder sonography as primary surveys. Patients with flank pain and hydronephrosis but without definitive radiopaque lesion received computed tomography (CT) scan as an advanced intervention. Cystoscopy was ordered by the physician if the patient had gross hematuria and was suspected to have intravesical lesion. We correlated the urine nitrate, the median of red blood cells (RBCs) and white blood cells (WBCs), and the ratio of red and WBCs with the finding on image studies, which included sonography, X-ray, and CT by statistical models. Results: In patients with urolithiasis detected by either X-ray or CT, logistic regression showed that higher urine RBC and lower urine WBC medians were predictors of urolithiasis. In the ROC analysis, WBC median lower than 7.75 per high power field (HPF) was a significant predicting factor for the positive finding of urolithiasis on a CT scan. Conclusion: When presenting with acute UTI symptoms, patients with a lower urine WBC median are correlated with a higher probability to have urolithiasis. In addition, WBC median lower than 7.75 per HPF in urinalysis is a predictor for urolithiasis found on CT in patients with flank pain or hydronephrosis.
Keywords: Computed tomography, urinary tract infection, urine marker, urolithiasis
How to cite this article:
Chen Y, Liao CH, Chiang BJ, Huang HC, Chou YW. A lower urine white blood cell median can be a predictor of undiscovered urolithiasis in patients with acute urinary tract symptoms. Urol Sci 2020;31:115-21 |
How to cite this URL:
Chen Y, Liao CH, Chiang BJ, Huang HC, Chou YW. A lower urine white blood cell median can be a predictor of undiscovered urolithiasis in patients with acute urinary tract symptoms. Urol Sci [serial online] 2020 [cited 2023 May 28];31:115-21. Available from: https://www.e-urol-sci.com/text.asp?2020/31/3/115/287973 |
| Introduction | |  |
Acute urinary tract infection (UTI) symptoms, as defined by the International Continence Society, include increased bladder sensation, urgency, frequency, dysuria, and pain in the lower urinary tract.[1],[2] These aforementioned symptoms are common. Two population-based survey studies have demonstrated that 60%–70% of men and women reported experiencing some degree of these symptoms.[3],[4]
The underlying causes of the symptom are diverse, including a UTI, inflammation, drug, lower urinary tract obstruction, iatrogenic injury, neurogenic diseases, and malignancy.[1] Therefore, further laboratory or image surveys may be needed for differential diagnosis. Urine tests, including urinalysis and urine culture, are common and convenient as a primary intervention for acute UTI symptoms. However, there are still some limitations to urine tests.
Pyuria, which is detected through microscopic evaluation of urine, is one of the most sensitive parameters for differentiating the cause of acute urinary infection symptoms.[5] Although a white blood cell (WBC) count ≥10 per high power field (HPF) is universally employed as a strong suggestion to diagnose UTI, its positive predictive value was as low as 40%.[5] Acute UTI can show symptoms such as urolithiasis, tumor, or other foreign material, which are not always easily differentiated from a real UTI.[1],[6],[7] Hematuria is commonly seen in patients with a UTI, urolithiasis, or urological neoplasm, usually requiring further investigation. In patients with symptomatic hematuria, there are chances of harboring an underlying disease, although symptomatic hematuria by itself does not justify routine image intervention.[7]
Urine culture is the gold standard to diagnose UTI. However, urine culture usually requires 5 days until the final report can be generated, which warrants another visit to the outpatient clinic.[8],[9] Thus, developing a predictor from parameters of urinalysis could help clinicians to order intervention just in time and avoid missing the underlying causes of the symptoms.[10] In this study, we aimed to identify predictive parameters from the urinalysis to detect underlying causes other than UTI in patients with acute UTI symptoms. These predictors could prompt further urological investigation and can be easily checked in an outpatient clinic.
| Materials and Methods | | |
We analyzed the data of 192 patients who presented with acute UTI symptoms in outpatient clinics from a single institute (Cardinal Tien Hospital, New Taipei City, Taiwan) between September 2012 and September 2018.
All patients with acute UTI symptoms underwent urinalysis. In addition, renal sonography and urinary bladder sonography were performed during the clinic visit. Each patient received a kidney–ureter–bladder (KUB) plain film X-ray. Patients with flank pain or hydronephrosis but without obvious radiopaque lesion received a computed tomography (CT) scan due to the suspicion of upper urinary tract lesions. Patients with gross hematuria or suspicious intravesical lesions found by the image studies received cystoscopy.
Patients with preexisting urological malignancy, urological congenital anomalies, and untreated urolithiasis were excluded. Initial symptoms, including frequency, urgency, dysuria, and pain, in the lower urinary tract (including lower abdominal pain and suprapubic pain) were recorded. Furthermore, a history of diabetes mellitus was recorded.
A total of 10 ml midstream urine was collected and then centrifuged. The supernatant fluid was decanted, and the urine sediment was resuspended. Subsequently, one drop of the urine sediment was poured onto a glass slide and covered for examination. The red blood cell (RBC)/WBC count was examined at 10 visual fields at HPF (400X) and reported as a range. The median value was used to represent RBC and WBC counts for statistical analysis (recorded as RBC median and WBC median thereafter). The ratio of the RBC median to the WBC median was also calculated and recorded as the RBC/WBC ratio. Renal and urinary bladder sonographies were performed using a 3.75 MHz probe (NemioXG, Toshiba, Japan) by a trained urologist during the clinic visit under the sensation of bladder fullness. Data were collected retrospectively. Chi-square test and Fisher's exact test were conducted for categorical variables. A logistic regression, with the calculation of the coefficient β and 95% confidence interval, was performed to detect the independent predictor of the positive finding on sonography, X-ray, and CT scan. ROC analysis with an AUC curve was utilized to analyze the diagnostic ability of urine markers for positive findings on an imaging study. We set the significance level at a two-sided P < 0.05. All data analyses were performed using SPSS, version 17.0, statistical software (SPSS Inc., Chicago, IL, USA).
The study was approved by the institutional review board of Cardinal Tien Hospital. The review board exempted the study from individual inform consent due to the retrospective and observational design of our study. IRB number: CTH-107-3-5-031, 2019/1/10.
| Results | | |
The demographic characteristics of included patients and the finding on sonography and KUB X-ray were summarized in [Table 1]. During the study period, 192 patients were included. The mean (±standard deviation) age was 55.23 ± 16.24 years. A total of 107 (55.7%) patients were men. Most patients were not diabetic (79.1%). Dysuria was the complaint of by 92 (47.9%) patients, which was the most common lower urinary tract symptom, followed by pain in the lower urinary tract and frequency. Only 18 (9.4%) patients experienced urgency. Forty-seven (24.5%) patients had lesions detected by sonography, which included 31 with urolithiasis (16.1%) and 18 (9.4%) with hydronephrosis. 65 (33.9%) patients were found to have urolithiasis by an X-ray. 81 patients underwent CT upon suspicion of urinary tract lesions or hydronephrosis, and 18 (22.2%) of the 81 had positive findings. Of those included subjects, 22 patients received cystoscopy due to either gross hematuria or suspicion of bladder lesion. Among them, 2 were found to have bladder cancer, 2 had benign bladder tumor, and 2 had bladder stone.
There was a trend, though not statistically significant, of higher RBC median and lower WBC median in the group with urolithiasis discovered by either X-ray or CT compared to those without urolithiasis [Table 1] and [Table 2]. Therefore, binary logistic regression was performed to detect the predictors of urolithiasis or tumor lesions from the urine.
[Table 3]a and [Table 3]b list binary logistic regression analysis results for the positive finding on imaging studies after adjustment for age and diabetes. The model in [Table 3]a included urine RBC median, urine WBC median, and urine nitrate. The model in [Table 3]b included urine RBC/WBC ratio and urine nitrate. The presence of urinary tract stones on X-ray and CT scan was significantly related to the median of urine RBC and WBC. A higher RBC median and a lower WBC median were significantly associated with a positive finding of urolithiasis on X-ray and CT scan, respectively. There was a similar trend in the relationship between urine RBC/WBC ratio and the KUB and CT findings on both t-test and logistic regression, despite not reaching a statistical significance.
In the ROC analysis, only a high WBC median was a significant predicting factor for the negative finding of urinary tract stones on a CT scan. The area under the curve was 0.77 [P< 0.01; [Figure 1]. [Figure 1] and [Figure 2] demonstrated the ROC analysis for urine RBC medium, urine WBC medium, and urine RBC/WBC ratio for predicting positive finding on CT scan or KUB X-ray examination. Using a WBC median of 7.75 as a cutoff value, Youden's index in the ROC analysis higher than the value carries a sensitivity of 73% and a specificity of 72.2% on predicting negative stone found on a CT scan. The urine RBC median and urine RBC/WBC ratio showed no significant correlation finding of urinary tract stones on both X-ray and CT scans. | Figure 1: Receiver operating characteristic curve of the relationship between specificity and sensitivity for positive findings on KUB X-ray of (a) urine RBC median, area under curve = 0.546, P =0.296; (b) urine white blood cell median, AUC = 0.537, P = 0.399; (c) urine RBC/white blood cell ratio, AUC = 0.546, P = 0.297
Click here to view |
 | Figure 2: Receiver operating characteristic curve of the relationship between specificity and sensitivity for positive findings on computed tomography of (a) urine RBC median, area under curve = 0.455, P = 0.566; (b) urine white blood cell median, AUC = 0.77, P = 0.001; (c) urine RBC/white blood cell ratio, AUC = 0.637, P = 0.077
Click here to view |
| Discussion | | |
Acute UTI symptoms are often noticed in the general population. In a population-based survey study, frequency and urgency were reported to be experienced by 6.8% and 10.8% of men and 7.4% and 12.4% of women, respectively.[3] Patients with UTI often experience these symptoms. UTI is not uncommon, particularly in women, in whom the lifetime risk of UTI is more than 50%.[11],[12] However, not all patients with acute UTI symptoms have a UTI. In their observational cohort study, Tomas et al. reported that, among 264 patients who visited the emergency department presenting with lower urinary tract symptoms, 175 (66%) were diagnosed as having UTI, but of patients for whom urine cultures were obtained, only 84 (48%) were found to be positive. The results suggest that overdiagnosis of UTI is common in patients with acute UTI symptoms leading to unnecessary use of antibiotics.[13]
In patients with acute UTI symptoms, identifying the underlying cause of symptoms is crucial. However, a negative urine culture usually takes 5 days to be reported, requiring a second clinic visit. Therefore, at an outpatient clinic, an oral antibiotic is often prescribed as an empirical treatment for these patients.[14] Moreover, empirical antibiotic treatment for a UTI usually takes 5–10 days which may delay the definitive diagnosis of the underlying causes if those symptoms were not caused by a UTI. Hence, it is of clinical importance to develop a convenient predictor for differentiating causes for the acute UTI symptoms.
Urolithiasis can also lead to acute UTI symptoms.[15] Recently, studies have been conducted to discover a marker for predicting urolithiasis. Porowski et al. obtained urine samples from pediatric stone formers and healthy controls. They found that the upper metastable limit osmolality can predict stone formation. However, it requires a 24-hour urine collection, which is not convenient to perform at a clinic visit.[16]
Okada et al. identified interleukin (IL)-4, IL-1a, granulocyte-macrophage colony-stimulating factor, IL-1b, and IL-10 as urinary inflammation-related markers that could distinguish stone formers from non-stone formers. These inflammatory markers could be released from macrophages when reacting to stones in the urinary tract.[17] Wang et al. identified metabolites of calcium oxalate stones through chromatography. Inclusion of products of caffeine, phenylalanine, galactose, and tyrosine metabolism showed a promising predicting ability to discriminate calcium oxalate stones former from healthy individuals.[10] Nonetheless, the identification of both inflammatory markers and metabolites from calcium compounds requires a specialized lab and considerable time to complete the tests. Therefore, the current role of these markers is to predict the risk of stone formation but not to differentiate the existence of urolithiasis.
Apart from aiming to discover markers for stone formation, studies have been conducted to determine convenient predictors for stone passages. Elmacı et al. reported that, in addition to smaller stone size, presence of microscopic or macroscopic hematuria was an independent predictor of stone passage.[18] In contrast, Lee et al. reported that a low neutrophil-to-lymphocyte count in blood was a predicting factor for spontaneous stone passage. They did not find markers in the urine as predictors. The low neutrophil-to-lymphocyte count might indicate a lowered inflammatory response due to the passage of a stone.[19]
In the present study, in patients with acute UTI symptoms, the urine RBC and WBC median could predict positive findings on X-ray and CT scans but not on sonography through logistic regression. Sonography is a technician-dependent study, and its result is limited by the patient's ability to cooperate with the practitioner; furthermore, bowel gas and other patient conditions can affect the results of the study. X-ray and CT scans are not affected by the technician's experience; therefore, X-ray and CT scan reports can be more objective compared with sonography.
Studies have been conducted to compare the ability of different imaging modalities in investigating hematuria. In a study by Ahmed et al., patients with suspected lower urinary tract-related hematuria underwent sonography before receiving cystoscopy. They found that transabdominal sonography had a sensitivity of 64.7% and 15.3% in detecting bladder stone and bladder tumor, respectively; therefore, cystoscopy has remained the gold standard for diagnosis.[20] Tan et al. compared renal and bladder ultrasound with CT. They found that ultrasound was poor at identifying renal calculi and urothelial tumors. In patients with hematuria and high suspicion of upper urinary tract disease, CTU should be considered.[21] Rajaie compared sonography and intravenous urography and found that ultrasound had a higher sensitivity in detecting urolithiasis and urinary tract tumors, and the author was in favor of using ultrasound in the initial evaluation of hematuria. However, the patient enrollment criteria did not include symptoms.[22]
The present study focused on patients visiting a clinic for acute UTI symptoms. Diament reported that performing screening sonography in all patients with suspected UTI was not cost-effective.[23] In addition, Bosmans et al. reported that, in women presenting with lower urinary tract symptoms, such as frequency or painful voiding, who contacted their general practitioner, performing serial examination rather than multiple tests was more cost-effective.[24] Hence, a predictive urine parameter could help clinicians to select the patients who need further intervention from patients with acute UTI symptoms.
The urine RBC median to urine WBC median ratio was included in the regression model because we hypothesize it could be a predictor of the existence of a bleeder. Fang et al. reported that a WBC to RBC cell count ratio in the diagnostic peritoneal lavage of ≥1 had a specificity of 97% and a sensitivity of 100% in predicting internal bleeding due to hollow organ perforation.[25] However, in the logistic regression model, the urinary cell count ratio was not able to reach a significant association with the positive findings in image studies.
According to our study, in patients with acute UTI symptoms, a higher RBC median and lower WBC median correlate with a positive image finding of urolithiasis on X-ray and CT scan. In patients who presented with flank pain or hydronephrosis, a urine WBC count median of <7.75 per HPF suggests a higher possibility of positive findings of urolithiasis on a CT scan. Based on our results, a urinalysis test for patients with acute UTI symptoms could provide more information to clinicians in terms of making decisions about conducting further imaging studies.
Twenty-two cases received cystoscopy exam and 2 later proved to have bladder cancer, which were also found by the sonography exam before cystoscopy. Indicating that sonography if performed properly may be a quick and sensitive image modality for cases with gross hematuria. However, two cases with bladder stones were detected with a cystoscopy exam and not by a bladder sonography. Cystoscopy exam still cannot be totally replaced with image studies. Furthermore, two cases with bladder cancer had a lower urine WBC median than the RBC median. Future studies enrolling more cases with bladder cancer might be done to establish the ability of urine RBC or WBC to predict bladder malignancy in patients with acute UTI symptoms.
The present study has several limitations. First, this retrospective study had inevitable selection bias and the small number of patients might have influenced the results. Second, only half of our patients received CT, which may cause a selection bias. Not all patients included with predominant urine RBC received cystoscopy exam. Bladder malignancy may also present as acute UTI symptoms. Early bladder cancer might not be detected by image study such as sonography or CT scan. Cases of bladder cancer might be missed without cystoscopy. Additional studies with a larger number of patients including cystoscopy exam are required to further determine the clinical significance of the urine RBC and WBC median in predicting the emergence of urolithiasis even cancer, to aid clinicians in decision-making regarding ordering an imaging study. Lastly, because imaging studies were reviewed by different radiologists, interpersonal differences could not be avoided. Nevertheless, imaging studies were reviewed by experienced radiologists; thus, the bias should be limited.
| Conclusion | | |
In summary, in patients presenting with acute UTI symptoms, a lower WBC median and a higher RBC median in urinalysis are correlated with a higher probability to have urolithiasis.
Furthermore, in patients with flank pain or hydronephrosis, a urine WBC median of <7.75 per HPF has a sensitivity of 73% for predicting urolithiasis on a CT scan. These findings provide information to clinicians for differential diagnosis and treatment in patients with acute UTI symptoms.
Declaration
The study was approved by the institutional review board of Cardinal Tien Hospital. The review board exempted the study from individual inform consent due to the retrospective and observational design of our study. IRB number: CTH-107-3-5-031.
Financial support and sponsorship
This research was financially supported by Cardinal Tien Hospital, New Taipei City, Taiwan (Grant no. CTH108A-2A25 and CTH108A-2A26).
Conflicts of interest
There are no conflicts of interest.
| References | | |
| 1. | Guralnick ML, O'Connor RC, See WA. Assessment and management of irritative voiding symptoms. Med Clin 2011;95:121-7.  |
| 2. | D'Ancona C, Haylen B, Oelke M, Abranches-Monteiro L, Arnold E, Goldman H, et al. The international continence society (ICS) report on the terminology for adult male lower urinary tract and pelvic floor symptoms and dysfunction. Neurourol Urodyn 2019;38:433-77. |
| 3. | Irwin DE, Milsom I, Hunskaar S, Reilly K, Kopp Z, Herschorn S, et al. Population-based survey of urinary incontinence, overactive bladder, and other lower urinary tract symptoms in five countries: Results of the EPIC study. Eur Urol 2006;50:1306-15. |
| 4. | Coyne KS, Sexton CC, Thompson CL, Milsom I, Irwin D, Kopp ZS, et al. The prevalence of lower urinary tract symptoms (LUTS) in the USA, the UK and Sweden: Results from the Epidemiology of LUTS (EpiLUTS) study. BJU Int 2009;104:352-60. |
| 5. | Kupelian AS, Horsley H, Khasriya R, Amussah RT, Badiani R, Courtney AM, et al. Discrediting microscopic pyuria and leucocyte esterase as diagnostic surrogates for infection in patients with lower urinary tract symptoms: Results from a clinical and laboratory evaluation. BJU Int 2013;112:231-8. |
| 6. | Torricelli FC, Chueh SC, Shen S, Monga M. Multiple uric acid bladder stones: Clinical presentation and endoscopic management. J Endourol Case Rep 2017;3:21-3. |
| 7. | Zhou Y, van Melle M, Singh H, Hamilton W, Lyratzopoulos G, Walter FM. Quality of the diagnostic process in patients presenting with symptoms suggestive of bladder or kidney cancer: A systematic review. BMJ Open 2019;9:e029143. |
| 8. | Price TK, Dune T, Hilt EE, Thomas-White KJ, Kliethermes S, Brincat C, et al. The clinical urine culture: Enhanced techniques improve detection of clinically relevant microorganisms. J Clin Microbiol 2016;54:1216. |
| 9. | Brubaker L, Wolfe AJ. The female urinary microbiota/microbiome: Clinical and research implications. Rambam Maimonides Med J 2017;8:e0015. |
| 10. | Wang X, Wang M, Ruan J, Zhao S, Xiao J, Tian Y. Identification of urine biomarkers for calcium-oxalate urolithiasis in adults based on UPLC-Q-TOF/MS. J Chromatogr B Analyt Technol Biomed Life Sci 2019;1124:290-7. |
| 11. | Griebling TL. Urologic diseases in america project: Trends in resource use for urinary tract infections in women. J Urol 2005;173:1281-7. |
| 12. | Foxman B. Recurring urinary tract infection: Incidence and risk factors. Am J Public Health 1990;80:331-3. |
| 13. | Tomas ME, Getman D, Donskey CJ, Hecker MT. Overdiagnosis of urinary tract infection and underdiagnosis of sexually transmitted infection in adult women presenting to an emergency department. J Clin Microbiol 2015;53:2686-92. |
| 14. | Leber AL, Thomas S. Alexander, Vickie S. Baselski, Deirdre L. et al. Clinical Microbiology Procedures Handbook. 4th ed., WILEY, 2016. |
| 15. | Lowe FC, Michel MC, Wruck JM, Verbeek AE. Is dipstick urinalysis screening beneficial in men with lower urinary tract symptoms? Adv Ther 2019;36:2954-67. |
| 16. | Porowski T, Kirejczyk JK, Mrozek P, Protas P, Kozerska A, Łabieniec Ł, et al. Upper metastable limit osmolality of urine as a predictor of kidney stone formation in children. Urolithiasis 2019;47:155-63. |
| 17. | Okada A, Ando R, Taguchi K, Hamamoto S, Unno R, Sugino T, et al. Identification of new urinary risk markers for urinary stones using a logistic model and multinomial logit model. Clin Exp Nephrol 2019;23:710-6. |
| 18. | Elmacı AM, Dönmez Mİ, Akın F, Çetin B, Gündüz M. What predicts spontaneous passage of ≤1cm ureteral stones in children? J Pediatr Surg 2019;S0022-3468(19)30360-4. |
| 19. | Lee KS, Ha JS, Koo KC. Significance of neutrophil-to-lymphocyte ratio as a novel indicator of spontaneous ureter stone passage. Yonsei Med J 2017;58:988-93. |
| 20. | Ahmed FO, Hamdan HZ, Abdelgalil HB, Sharfi AA. A comparison between transabdominal ultrasonographic and cystourethroscopy findings in adult Sudanese patients presenting with haematuria. Int Urol Nephrol 2014;47:223-8. |
| 21. | Tan WS, Sarpong R, Khetrapal P, Rodney S, Mostafid H, Cresswell J, et al. Can renal and bladder ultrasound replace computerized tomography urogram in patients investigated for microscopic hematuria? J Urol 2018;200:973-80. |
| 22. | Esfahani MR, Momeni A. Comparison of ultrasonography and intravenous urography in the screening and diagnosis of hematuria causes. Urol J 2006;3:54-60. |
| 23. | Diament MJ. Is ultrasound screening for urinary tract infection “cost effective?” Pediatr Radiol 1998;18:157-9. |
| 24. | Bosmans JE, Coupé VM, Knottnerus BJ, Geerlings SE, van Charante EP, ter Riet G. Cost-effectiveness of different strategies for diagnosis of uncomplicated urinary tract infections in women presenting in primary care. PLoS One 2017;12:e0188818. |
| 25. | Fang JF, Chen RJ, Lin BC. Cell count ratio: New criterion of diagnostic peritoneal lavage for detection of hollow organ perforation. J Trauma Acute Care Surg 1998;45:540-4. |
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]
|
Search Pubmed for