Article Text

Download PDFPDF

Nitrate in community water supplies and incidence of non-Hodgkin’s lymphoma in Sardinia, Italy
  1. P Cocco1,
  2. G Broccia2,
  3. G Aru1,
  4. P Casula2,
  5. S Muntoni3,
  6. K P Cantor4,
  7. M H Ward4
  1. 1Department of Public Health, Occupational Health, University of Cagliari, Italy
  2. 2Division of Haematology, Oncology Hospital, ASL 8, Cagliari, Italy
  3. 3Department of Biomedical Sciences and Biotechnologies, University of Cagliari
  4. 4Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, USA
  1. Correspondence to:
 Dr P Cocco, Department of Public Health, Occupational Health Section, University of Cagliari, via San Giorgio 12, 09124 Cagliari, Italy; 

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Studies of non-Hodgkin’s lymphoma (NHL) evaluating risk associated with nitrate concentrations in community water supplies have found inconsistent results.1–4 To further explore this hypothesis, we compared the NHL incidence in 1974–1993 with nitrate monitoring data from 1971–1994 available for 75% (282 of 376) of the communes (administrative units) in Sardinia, Italy.


We selected the 153 communes with at least one nitrate measurement in 1993 or earlier (median: 5 measurements; range: 1–51) after excluding the six major urban areas. The population of these 153 communes ranged from 140 to 61 636 (median: 2459) and included a population of 703 000 (41% of the Sardinian population). There were a total of 3909 nitrate measurements. The nitrate ion was measured by spectrophotometry with the sodium salicilate method. For our primary analysis, we used the average nitrate concentration in 1993 as the exposure metric. Fifteen communes (23% of the study population), also had nitrate data in 1971–84. For these we also evaluated the median concentration in 1971–84. NHL incidence in 1974–1993 was determined by commune from the registries of all pathology and haematology departments in Sardinia. Coverage of these registries was estimated as close to 100%.5 We calculated person years for each gender group and 10 year age group in each commune, using the 1971, 1981, and 1991 census enumerations, extended five years backward and four years onward. We aggregated cases and person years over all communes within eight categories of 1993 nitrate level. Incidence rate ratios (IRR) and 95% confidence intervals for NHL were computed for each nitrate category with reference to the lowest using Poisson regression, and adjusting by gender, age, and population size.


Among the 153 study communes, the average nitrate concentration in 1993 was 4.57 mg/l (SE 0.35; median: 3.27). The IRRs for NHL for both genders combined did not increase with increasing 1993 nitrate level (table 1). Among men, the IRRs were significantly increased in some nitrate concentration categories, but no linear increase was observed (χ2 for trend=3.25: 0.10>p>0.05). Among women, the IRRs were not increased in any exposure category. The results did not change using different exposure categories cut off points, and results were similar for the NHL incidence in 1974–83 and in 1984–93. Among the 15 communes with monitoring data in 1971–84, the median nitrate concentration was 8.11 mg/l (IQR 4.42–24.34), and it was significantly correlated with the 1985–1994 median (r=0.577; p<0.05), and with the 1993 level (r=0.613; p<0.05). Nitrate concentrations in 1993 were substantially lower in these 15 communes (median: 4.10; IQR 2.91–8.06) compared with the 1971–84 median concentration, and the IRRs for NHL were increaseed 2-fold to 2.5-fold for increasing quartile of nitrate concentrations compared with the lowest. The risk increase was consistent by gender. Still, no significant upward trend was observed. However, the incidence rate in the reference category was based only on seven cases and 116 778 person years and when the reference category was extended to <8.0 mg/l nitrate, the IRR was 1.3 for 8–24.9 mg/l, and 1.0 for ⩾25 mg/l. Finally, we repeated the analyses in table 1 after excluding these 15 communes. This resulted in a reduction of the raised IRRs for NHL. Only one AIDS related NHL case occurred among the 737 NHL cases observed in our study population. Therefore, we can reasonably exclude that the urban/rural gradient of AIDS related NHL might have significantly affected our findings.

Table 1

Relative risks (and 95% confidence intervals) for non-Hodgkin’s lymphoma and 1993 nitrate concentration in community water supplies for the total population and by sex


We found limited evidence among men but not women of an association between NHL incidence and nitrate concentrations in community water supplies. Low average concentrations, exposure misclassification, and the limitations of the ecological design might account for our results. Further studies with historical monitoring data, individual estimates of exposure, and information on potential confounders areneeded to adequately evaluate the hypothesis that NHL risk is associated with increased nitrate concentrations in drinking water.



  • Funding: none.

  • Conflicts of interest: none.