Skip to main content

Typhoid Fever From Water Desalinized Using Reverse Osmosis*.


In May 1992, officials in Tabuk City (population 130,000) in northwestern Saudi Arabia became concerned about increasing reports of typhoid fever (TF), including several among staff of the city's referral hospital (KKH). Since no water source in the city was chlorinated, we first enforced chlorination of all water, including water used for 20 private reverse osmosis (RO) drinking water treatment plants, 10 municipal wells and five private wells. We then initiated an investigation to identify the source and method of transmission.


We required all Tabuk medical facilities to report suspected TF and provided additional laboratory support for blood, stool and urine cultures. We defined confirmed TF as suspicion of TF by a physician with Salmonella typhi isolated from blood, stool or urine. We compared household water sources of each TF cases with those of 199 control houses selected at random from all houses in Tabuk City.


We identified 81 confirmed TF cases, including nine in KKH staff (Attack Rate [AR] 140/10,000), two in family of staff, 57 in the community (AR 4.4/10,000) and 13 in a local military cantonment. The outbreak began with the onset of TF in all three areas within five days of one another, continued for seven weeks, and ended two weeks after chlorination began. TF was found in all districts of Tabuk, with AR ranging from 0.9 to 10.3 per 10,000. Among water sources the odds ratio was highest (2.5; 95% CI 1.2, 5.4) for water purchased from RO plants. Water purchased from RO plants supplied by one well (Well A) carried the highest risk (OR=7.1; 95% CI 2.5, 20), while other RO source wells had OR near or less than 1.0. During the outbreak the KKH well had failed and for one month the hospital had obtained water for RO (300,000 liters per day) from Well A. The aquifer for Well A lay partially beneath a depression where city sewage collected. Water (unchlorinated) sampled from Well A one month after the outbreak ended yielded coliforms.


Water for RO must be prechlorinated to prevent microbiologic fouling of the membranes. Well A probably became contaminated with S. typhi when KKH demand overtaxed the aquifer and drew in surface water. Membranes in RO plants using this unchlorinated well water would then become fouled with S. typhi. Private RO plants are common throughout Saudi Arabia and will need more intensive scrutiny of their operation and water purity.