Response: Re: Population-Based Case–Control Study of Chinese Herbal Products Containing Aristolochic Acid and Urinary Tract Cancer Risk

(1)

988 Correspondence | JNCI Vol. 102, Issue 13 | July 7, 2010

CorrespondenCe

response

Su et al. express concern about the sample representativeness and the comparability of age and sex distributions between the case and control groups in our study (1). We included all newly diagnosed patients with urinary tract cancer in Taiwan for the case patients, and the control subjects (n = 174 701) were selected from a simple ran-dom sampling of the general population; thus, the study sample was representative of the general population of Taiwan. Because we adjusted for age and sex in the logistic regression model, the compara-bility within individual strata was assured. Because matching in case–control studies may produce selection bias and must be combined with a stratified analysis to

at National Taiwan Univ. on October 13, 2010

jnci.oxfordjournals.org

(2)

jnci.oxfordjournals.org JNCI | Correspondence 989 Table 1. Frequency distributions of various risk factors and adjusted odds ratios (ORs)

with 95% confidence intervals (CIs) for new occurrence of urinary tract cancer in patients older than 50 years from multivariable logistic regression model*

Risk factor Case subjects (N = 3954) Control subjects (N = 31 081) Adjusted OR (95% CI)† P Sex Female 1371 14 150 1.0 (referent) Male 2583 16 931 1.6 (1.5 to 1.7) <.001 Age, y 50–59 753 14 429 1.0 (referent) <.001 60–74 1932 11 308 3.2 (3.0 to 3.5) <.001 75–99 1269 5344 4.6 (4.2 to 5.0) <.001 Residence in township where black foot disease was endemic

No 3871 30 948 1.0 (referent)

Yes 83 133 4.6 (3.5 to 6.2) <.001

Chronic UTI

No 3873 30 708 1.0 (referent)

Yes 81 373 1.7 (1.3 to 2.1) <.001

Mu Tong, total amount prescribed, g 0 3440 27 069 1.0 (referent) 1–60 416 3512 1.1 (0.9 to 1.2) .191 61–100 42 243 1.7 (1.2 to 2.5) .003 101–200 38 178 2.1 (1.4 to 3.1) <.001 >200 18 79 2.2 (1.2 to 3.8) .007 Each 30-g increase‡ NA NA 1.1 (1.07 to 1.2) <.001

Fangchi, total amount prescribed, g

0 3377 25 851 1.0 (referent)

1–60 535 4934 0.9 (0.8 to 1.0) .232

61–100 15 148 0.7 (0.4 to 1.3) .282

>100 27 148 1.3 (0.9 to 2.1) .212

Xi-Xin, total amount prescribed, g

0 3166 24 741 1.0 (referent)

1–100 723 5823 1.1 (1.01 to 1.2) .029

101–300 48 420 0.8 (0.6 to 1.2) .294

>300 17 97 1.0 (0.6 to 1.8) .955

* NA = not applicable; UTI = urinary tract infection.

† Adjusted for age, sex, residence in a township where black foot disease was endemic, and history of chronic UTI with logistic regression model for different dosages of Chinese herbs.

‡ Estimation of odds ratio based on continuous variable for every 30-g increment of Mu Tong.

improve efficiency, we chose to construct a multivariable logistic regression model (2). In fact, both age and sex were previously found to be associated with the frequency of having been prescribed Chinese herbal products (3). We reanalyzed the data by limiting both case patients and control subjects to those who were older than 50 years and obtained similar results (Table 1).

Su et al. suggested that we include the use of acetaminophen and nonsteroidal anti-inflammatory drugs in the analysis. We have conducted such an analysis and obtained results that were similar to those of our original analysis (Mu Tong: at 61– 100 g, odds ratio = 1.4, 95% confidence

interval = 1.1 to 1.9; at >200 g, odds ratio = 1.7, 95% confidence interval = 1.1 to 2.7). In fact, the association between analgesic consumption and increased risk of urinary tract or bladder cancer might be controver-sial (4,5). To avoid misinterpretation of potential interactions among different risk factors, we deliberately excluded partici-pants who were prescribed more than 500 pills of analgesics.

In our study, more than 100 g of Fangchi was statistically significantly associated with an increased crude odds ratio for urinary tract cancer; however, the association did not reach statistical significance after ad-justment for other risk factors, probably

because of the small number of case sub-jects. In the estimation of total dose, we accumulated the prescribed doses for case subjects up to 1 year before diagnosis, whereas the prescribed doses for control subjects were accumulated to the end of 2002, which might have increased misclas-sification and underestimated the effect.

Finally, Su et al. comment that we should have controlled for comorbid con-ditions and the medications used to treat them. Because a confounder must be a risk predictor of the outcome (2), only medica-tions that are reported to be associated with urinary tract cancer [eg, cyclophosphamide (6)] can be a potential confounder, whereas indications for being prescribed Mu Tong that have never been reported to poten-tially cause urinary tract cancer (eg, chronic hepatitis, rhinitis, dysmenorrhea, or ec-zema) cannot produce confounding in this study. even though no case subject in our study was ever prescribed cyclophosphamide before the development of the cancer and even though we controlled for arsenic expo-sure and chronic urinary tract infection and partially controlled for smoking (through the surrogate indicator of sex), we still found an association between having been prescribed Chinese herbal products containing aristo-lochic acid and the risk of urinary tract can-cer. Given the potential underestimation of the effect as mentioned above, we concluded that such an association exists.

JuNG-DeR WANG MING-NAN LAI PAu-CHuNG CHeN

references

1. Lai MN, Wang SM, Chen PC, Chen YY, Wang JD. Population-based case–control study of Chinese herbal products containing aristolochic acid and urinary tract cancer risk. J Natl Cancer Inst. 2010;102(3):179–186. 2. Rothman KJ, Greenland S, Lash TL. Design

strategies to improve study accuracy. In: Rothman KJ, Greenland S, Lash TL, eds. Modern Epidemiology. 3rd ed. Philadelphia, PA: Lippincott Williams and Wilkins; 2008: 171–182.

3. Hsieh SC, Lai JN, Lee CF, Hu FC, Tseng WL, Wang JD. The prescribing of Chinese herbal products in Taiwan: a cross- sectional analysis of the national health insur-ance reimbursement database. Pharmacoepidemiol Drug Saf. 2008;17(6):609–619.

4. Kaye JA, Myers MW, Jick H. Acetaminophen and the risk of renal and bladder cancer in the general practice research database. Epidemiology. 2001;12(6):690–694.

(3)

990 Correspondence | JNCI Vol. 102, Issue 13 | July 7, 2010 5. Castelao Je, Yuan J-M, Gago-Dominguez M,

et al. Non-steroidal anti-inflammatory drugs and bladder cancer prevention. Br J Cancer. 2000;82(7):1364–1369.

6. Travis LB, Curtis Re, Glimelius B, et al. Bladder and kidney cancer following cyclo-phosphamide therapy for non-Hodgkin’s lym-phoma. J Natl Cancer Inst. 1995;87(7):524–530. notes

Affiliations of authors: Institute of Occupational

Medicine and Industrial Hygiene, College of Public Health, National Taiwan University, Taipei, Taiwan (M-NL, P-CC, J-DW); Department of Internal Medicine and Department of Environmental and Occupational Medicine, National Taiwan University Hospital, Taipei, Taiwan (J-DW).

Correspondence to: Jung-Der Wang, MD, ScD,

Institute of Occupational Medicine and Industrial Hygiene, College of Public Health, National Taiwan University, Rm 719, No. 17 Xu-Zhou Rd, Taipei City 100, Taiwan (e-mail: [email protected] .tw).

DOI: 10.1093/jnci/djq181