Environmental health

To the Editor:We report an extensive investigation into a cluster of five confirmed cases of Legionnaires’ disease in Sydney’s Inner West in May 2016. The patients were aged 62–89 years, four were men, all were ex-smokers, and four had significant comorbidities, such as immunosuppression. Hospitalisations lasted a median of 6 days and one patient died. All patients tested positive for Legionella pneumophila serogroup 1 urinary antigens; however, L. pneumophila was isolated by culture in only one patient. Patients were interviewed to obtain the histories of where they had been in the 2–10 days before the onset of symptoms (the incubation period), and then, to identify the focus for our environmental investigation, we mapped the locations against water source locations.1 Local councils assisted in the collection of water samples: 86 samples were obtained from cooling towers and 15 from other sites, such as fountains. At the time of sampling, requests were made for immediate cleaning of all visibly unclean air conditioning cooling towers or those with cloudy water. Only one sample tested positive for Legionella: a fountain with a low positive result of Legionella anisa. If Legionella isolates are genetically indistinguishable, the genomic sequencing of clinical and environmental samples is a useful tool that can link cases to each other and to an environmental source. In this investigation, L. pneumophila was cultured in only one case and in no environmental samples; therefore, we could not link cases to each other or to an environmental source by genotyping, leaving the possibility that cases were unrelated and sporadic. Despite this investigation, no environmental source for these infections was identified. It is uncommon in an investigation of this size to not detect L. pneumophila in any cooling towers — a routine New South Wales survey found L. pneumophila in 2% of cooling towers in a non-outbreak situation.2 Possible explanations for not detecting Legionella include low sensitivity of sampling at one point in time, incidental and possibly unrelated cleaning of the source cooling tower before sampling, or underestimating the distance that contaminated aerosols may travel, meaning that the source cooling tower was not sampled.3 No further cases with similar exposures were reported in the period immediately after the environmental investigation. This suggests that the extensive public health interventions taken at the time to remedy the problems identified may have been effective, or that incidental cleaning of cooling towers before our sampling may have eliminated the potential source.

Marianne Dowsett · Emma Quinn · Leena Gupta

To the Editor:I question that clinicians and the public may be reassured that the evidence underpinning the renewal of the National Cervical Screening Program (NCSP) is sound.1 A fundamental aspect of the sensitivity of a screening test is the definition of the disease of interest that the test seeks to detect. For cancer screening, it is disease that will progress to life-threatening cancer and not the sensitivity to detect disease that only may progress to cancer.2,3 It is a very common but profound error to consider these equivalent,2 particularly when comparing two tests that measure different aspects of the disease process. Overdiagnosis and regression biases are avoided by using the interval cancer method to estimate sensitivity.2,3 The estimates of the sensitivity of human papillomavirus (HPV) testing and cytology, used in the models of the New South Wales group to justify the NCSP renewal policy, were for the detection of cervical intraepithelial neoplasia (CIN) grade 2+ and not for the detection of progressing disease.4 Their approach is fraught because CIN 2 and CIN 3 have different rates of regression and persistence. A higher sensitivity of HPV to detect CIN 2, but lower sensitivity to detect CIN 3 or progressive disease, will produce a lower effectiveness to reduce invasive cervical cancer than cytology, despite a higher sensitivity to detect CIN 2+. There is evidence from the randomised controlled trials that this is the case, even within the regression, persistence but no progression, and progression spectrum of CIN 3. The results of the only randomised trial of the implementation of HPV testing in a screening program, published in 2013,2 found the sensitivity of HPV testing (87%) to detect disease that progresses to invasive cervical cancer to be lower than cytology screening (93%). The authors concluded that “sensitivity of HPV testing was similar to that of Pap testing but caused more overdiagnosis. Therefore, implementation of HPV testing needs to be reconsidered especially in countries with well organised programmes”. This suggests that it is mathematically impossible for HPV testing to provide greater protection from invasive cervical cancer than cytology. Assuming participation of 80%, the renewal program will increase the incidence of cervical cancer by about 39%, and the unit of economic effectiveness will be dollars saved per life prematurely lost. From the perspective of public health medicine, it would be professionally negligent not to estimate the effect of moving to 5-yearly HPV testing using the Finnish estimates of the sensitivity of HPV testing and cytology, but instead rely only on estimates with major regression and overdiagnosis bias.4

Brian Cox

In reply

Jonathan Carter · Ian Hammond · Megan Smith

n/a

Simon Quilty · Thomas S Kaye · Bart J Currie

n/a

Satvinder S Dhaliwal · Timothy A Welborn

In reply

Rebecca McKetin · Louisa Degenhardt · Wayne D Hall