Send letter to journal:
Re: Infallibility of a normal platelet count/spleen diameter ratio in ruling out oesophageal varices?

Dear Editor

I read with interest that an abnormal platelet count/spleen diameter ratio predicts the presence of oesophageal varices.[1] This otherwise excellent article contained a statistical error that I would like to bring to your attention.

The authors report a 100% sensitivity of the diagnostic test (platelet count/spleen diameter ratio at a cutoff value of 909) in ruling out the diagnosis of esophageal varices: all patients with varices had an abnormal ratio. The reported 95% confidence interval (CI) for the sensitivity is 100-100%. This is simply incorrect, and is was probably calculated using a statistical formula that can be only used when:
1. the sample size is large, and
2. the proportion in question is not equal to 0% or 100%.

This formula:
(SE(p) = sqrt[p(1-p)/n]
where SE = standard error, p = proportion, and n = sample size)

gives a value for the standard error of a proportion. However, when the proportion is 0% or 100%, the standard error becomes 0, and this formula is void.

Imagine a diagnostic test that is performed on 10 patients, 2 of whom have the disease in question. Let us say that the test is positive in both patients with the disease and in 2 of the 8 patients without the disease. The calculated sensitivity is 100% and specificity is 75%. Would any respectable journal editor be convinced that the true sensitivity of this diagnostic test lies somewhere between 100% and 100%-- that if a validation cohort underwent this diagnostic test we could be confident that all patients with the disease should test positive? Of course not. But the above formula would give a standard error of 0 in this case as well.

The bottom line is that it is never acceptable to give a confidence interval in medicine that is a point estimate. If a computer program or statistical textbook formula gives a point estimate instead of an interval, then the wrong formula was used. This particular statistical error (expressing a confidence interval as a point estimate) is easily identified without any statistical training or calculations.

In the case of the platelet count/spleen diameter ratio, an appropriate statistical test is the score confidence interval (Agresti- Coull) method. Using this method (in JMP 5.1.1, SAS Institute, Cary, NC), the sensitivity of the platelet count/spleen diameter ratio in the derivation cohort should have been:
89 of 89 patients with varices correctly identified = 100% (95% confidence interval 95.9% to 100%)

and for the validation cohort:
71 of 71 patients = 100% (95% confidence interval 94.9% to 100%).

Combining the validation and derivation cohorts:
160 of 160 patients = 100% (95% CI 97.7% to 100%).

This is not nit-picky since many clinicians will read this article and believe that a normal platelet count/spleen size ratio rules out esophageal varices with 100% certainty. But surely someone, somewhere, will find an exception to this useful (and remarkably accurate) diagnostic rule.

References

1. E Giannini, F Botta, P Borro, D Risso, P Romagnoli, A Fasoli, M R Mele, E Testa, C Mansi, V Savarino, and R Testa. Platelet count/spleen diameter ratio: proposal and validation of a non-invasive parameter to predict the presence of oesophageal varices in patients with liver cirrhosis. Gut 2003; 52: 1200-1205

Send letter to journal:
Re: Platelet count/Spleen diameter ratio as a predictor of oesophageal varices in alcoholic cirrhosis.

Dear Editor

We read with great interest the article by E Giannini et al. (Gut 2003;52:1200-1205) regarding platelet count/spleen diameter ratio, as a non invasive parameter to predict the presence of oesophageal varices in patients with liver cirrhosis.[1] In patient with liver disease due to alcohol,platelet count is reduced due to the myelotoxic effect of alcohol.

In your study only 16,55% (24/145) of your patients had liver cirrhosis due to alcohol. In the United Kingdom, alcohol is the commonest cause of liver cirrhosis.

We retrospectively studied the Endoscopy, Haematology and Radiology reports of 40 patients who had been treated for alcohol induced cirrhosis at Homerton Hospital, London. Out of these 30 had oesophageal varices at endoscopy and 10 did not. The platelet count/spleen diameter ratio was calculated within two months of endoscopy.

The median platelet count / spleen diameter ratio in patients with varices was 537 (range 371-670) and with no varices was 2229(range 1542- 3174). A platelet count/spleen diameter ratio of < 909 had 100% sensitivity and specificity for the prediction of oesophageal varices in the patients.

We have shown that the non -invasive ways of predicting the presence of oesophageal varices through platelet count / spleen diameter ratio is reproducible in alcoholic cirrhotic patients.

Reference

1. Peck-Radosavljevic M. Thrombocytopenia in liver Disease.Can J Gastroenterol 2000;14 (suppl D):60-6D.

Send letter to journal:
Re: Platelet count with or without spleen diameter? The case when accompanied is better than alone.

Dear Editor

We were pleased to receive the interesting comments that Thabut et al. made to our paper.[1] Indeed, their letter allows us to focus on some aspects of our study that we feel need to be further emphasised.

As a general rule, a surrogate marker for a given variable (i.e., presence/absence of oesophageal varices) that already has a definite diagnostic means (i.e., endoscopy) should accomplish to two major criteria. Firstly, it should be the product of a thorough statistical analysis, and secondly, but no less importantly, it has to be biologically plausible.

From a statistical point of view, as Thabut et al. correctly point out, both the platelet count and the platelet count/spleen diameter ratio reached an excellent diagnostic accuracy for the non-invasive diagnosis of the presence/absence of oesophageal varices. However, we do not agree with their assumption that the use of the platelet count/spleen diameter ratio did not add significant discrimination to the use of the platelet count alone. In fact, the accuracy of the platelet count/spleen diameter ratio for the diagnosis of oesophageal varices is not only better than that of the platelet count alone, but is also significantly so. Briefly, in the cohort of 145 patients with compensated cirrhosis, which is the group that most likely benefits form screening, the difference between the AUC-ROC of the platelet count/spleen diameter ratio and the platelet count was 0.041 (0.013-0.070), with p=0.005 in favour of the platelet count/spleen diameter ratio. Moreover, in the whole cohort of 266 patients the platelet count/spleen diameter ratio had a c-index=0.902 (0.860-0.935, 95% confidence interval) while the platelet count alone had a c-index=0.839 (0.790-0.881), with a difference between AUC-ROC=0.063 (0.038-0.088) and p=0.001. Furthermore, the platelet count/spleen diameter ratio was the only parameter significantly associated to the presence/absence of oesophageal varices in a multivariate analysis that also included platelet count. Lastly, as recently underlined, the negative predictive power of a non-invasive parameter used to predict the absence/presence of oesophageal varices is a fundamental clinical concern.[2] In fact, for such a tool to be adopted in clinical practice it has to achieve a negative predictive value of 100%, though maintaining an acceptable positive predictive value. This would preserve the safety of the parameter (i.e. virtual absence of missing a diagnosis) and keep a satisfactory cost-efficacy profile. In practice, in our study the use of the platelet count/spleen diameter ratio proved to fulfil these criteria while the platelet count alone did not.

Biological plausibility is a not a secondary concern for the clinician. As we emphasised in our paper, as we recently demonstrated,[3] and as Thabut et al. also pointed out, the presence of thrombocytopenia in patients with liver cirrhosis is likely a multi-factorial event.[4] Therefore, the use of the platelet count to diagnose a feature that depends on portal hypertension alone may lead to an increase in false positive results, thus decreasing the accuracy as well as the cost- efficacy of the diagnostic means. As underscored in our paper, the use of the platelet count/spleen diameter ratio could by-pass this inconvenient by "normalising" the platelet count to the platelet count decrease effectively dependent on hypersplenism.

Lastly, some methodological issue should be taken into account when the two items of the ratio are singularly evaluated. On the one hand, the spleen diameter measurement should to be performed by a skilled operator, and its results should have excellent accuracy and reproducibility. On the other hand, we have shown that the consistency of the ratio is maintained even considering the expected, mild fluctuations in platelet count commonly seen in cirrhotic patients during a limited period of time.[1]

All in all, we did not have the presumption to propose a diagnostic "magic bullet" as today is more and more being desolately seen in clinical hepatology. We are well aware that the results we obtained have to be validated in independent series and/or in cohorts with different aetiology of liver disease before being widely accepted,[5] and in our paper we clearly stated the limitations of our study.[1] However, the population we dealt with is the population we commonly encounter in everyday clinical practice, and it is not so different from that seen in other parts of our Country, this meaning viral cirrhosis in approximately 70% of the patients.[6] Moreover, if we glance outside our borders we see that the population is not so different, viral aetiology of liver disease being the leading cause for liver transplantation in Europe during the period January 1998-December 2001 (22,924 cirrhotic patients).[7] Nevertheless, if we look back to our data we see that the use of the platelet count/spleen diameter ratio performs equally good in the limited subset of patients with alcoholic cirrhosis (n=53, platelet count spleen diameter ratio c-index=0.958, platelet count c-index=0.740, difference between AUC- ROC=0.218, p=0.001), although we feel that focusing on a specific sub- group of patients that does not reflect the true prevalence of the disease in the population would introduce a bias.

In conclusion, we proposed a new evaluation tool and actually called for validation of our method, being conscious that only difference of opinion that arises from results obtained in well-conducted studies contributes to scientific progress, and most importantly that "life is short, and Art long; the crisis fleeting; experience perilous, and decision difficult. The physician must not only be prepared to do what is right himself, but also to make the patient, the attendants, and externals cooperate."[8]

References

(1) Giannini E, Botta F, Borro, et al. Platelet count/spleen diameter ratio: proposal and validation of a non-invasive parameter to predict the presence of oesophageal varices in patients with liver cirrhosis. Gut 2003; 52: 1200-5.

(2) de Franchis R. Evaluation and follow-up of patients with cirrhosis and oesophageal varices. J Hepatol 2003; 38: 361-3.

(3) Giannini E, Botta F, Borro P, et al. Relationship between thrombopoietin serum levels and liver function in patients with chronic liver disease related to hepatitis C virus infection. Am J Gastroenterol 2003; 98: 2516-20.

(4) Peck-Radosavljevic M. Thrombocytopenia in liver disease. Can J Gastroenterol 2000; 14 (suppl. D): 60D-6D.

(5) Laupacis A, Sekar N, Stiell IG. Clinical prediction rules. A review and suggested modifications of methodological standards. JAMA 1997; 277: 488-94.

(6) De Bac C, Clementi C, Duca F, et al. Liver cirrhosis: epidemiological aspects in Italy. Res Virol 1997: 148: 139-42.

(7) European Liver Transplant Registry. Primary indications of Liver Transplantation in Cirrhosis in Europe. http://www.eltr.org/publi/results.php3?id_rubrique=48 Accessed November 26, 2003.

Send letter to journal:
Re: Prediction of oesophageal varices with platelet count/spleen diameter ratio or platelets alone ?

Dear Editor

We read with a great interest the article by Giannini et al. in a recent issue of Gut.[1]

Since the incidence of chronic liver diseases is growing, we are convinced that the development of non-invasive predictive tools to identify cirrhotic patients with oesophageal varices is of major interest. Several markers have been studied, and among them platelet count is commonly reported to be a good predictor of oesophageal varices. However, in the eight studies already published,[2-9] their discriminative power was moderate, with areas under the ROC curve of 0.70 or less, and this for platelets alone [5] and for indexes combinating platelets to other markers.[2,5,6] Most of these studies have included heterogeneous group of patients, with compensated and decompensated cirrhosis. In our unit, we performed prospectively platelet count and screening upper oesogastroduodenoscopy the same day to 41 patients with compensated cirrhosis and confirmed the moderate value of platelet count alone (AUROC=0.70 +/- 0.07) (Thabut, data not shown). The major drawback of platelet count is that it can depend on other factors than portal hypertension in cirrhotic patients. To avoid this bias, Giannini developed an index based on platelet count/spleen diameter ratio and found far better results than previous studies,[1] with a c index (equivalent to the area under ROC curve) of 0.92 for patients with compensated liver cirrhosis. However, we were surprised to see that the use of platelets/spleen diameter ratio does not add significant discrimination to platelet count alone (c index of 0.92 vs 0.88) in their population. Considering this point, their excellent results would not be explained by the discriminative power of their index but by the excellent diagnostic power carried by platelet count itself in their series. Several explanations can be raised, and one of them is the high rate of viral-related cirrhosis in their patients, where platelet count is less exposed to variations than in alcoholic patients for example. This point is of major concern for the further validation of their index, recommended by the authors themselves, in other populations. In conclusion, Giannini et al. have found a very performing index to predict the presence of oesophageal varices in cirrhotic patients. We think that the excellent performance they obtained was not due to their index but to the surprisingly good performance of platelet count alone in their patients. Considering the results carried by platelet count for the prediction of oesophageal varices in the already published studies, we fear that the warranted validation studies of this index show less exciting results.

References

(1) Giannini E, Botta F, Borro P, Risso D, Romagnoli P, Fasoli A, et al. Platelet count/spleen diameter ratio: proposal and validation of a non -invasive parameter to predict the presence of oesophageal varices in patients with liver cirrhosis. Gut 2003;52(8):1200-5.

(2) Chalasani N, Imperiale TF, Ismail A, Sood G, Carey M, Wilcox CM, et al. Predictors of large esophageal varices in patients with cirrhosis. Am J Gastroenterol 1999;94(11):3285-91.

(3) Madhotra R, Mulcahy HE, Willner I, Reuben A. Prediction of esophageal varices in patients with cirrhosis. J Clin Gastroenterol 2002;34(1):81-5.

(4) Ng FH, Wong SY, Loo CK, Lam KM, Lai CW, Cheng CS. Prediction of oesophagogastric varices in patients with liver cirrhosis. J Gastroenterol Hepatol 1999;14(8):785-90.

(5) Pilette C, Oberti F, Aube C, Rousselet MC, Bedossa P, Gallois Y, et al. Non-invasive diagnosis of esophageal varices in chronic liver diseases. J Hepatol 1999;31(5):867-73.

(6) Schepis F, Camma C, Niceforo D, Magnano A, Pallio S, Cinquegrani M, et al. Which patients with cirrhosis should undergo endoscopic screening for esophageal varices detection? Hepatology 2001;33(2):333-8.

(7) Thomopoulos KC, Labropoulou-Karatza C, Mimidis KP, Katsakoulis EC, Iconomou G, Nikolopoulou VN. Non-invasive predictors of the presence of large oesophageal varices in patients with cirrhosis. Dig Liver Dis 2003;35(7):473-8.

(8) Zaman A, Hapke R, Flora K, Rosen HR, Benner K. Factors predicting the presence of esophageal or gastric varices in patients with advanced liver disease. Am J Gastroenterol 1999;94(11):3292-6.

(9) Zaman A, Becker T, Lapidus J, Benner K. Risk factors for the presence of varices in cirrhotic patients without a history of variceal hemorrhage. Arch Intern Med 2001;161(21):2564-70.