SUMMARY STATEMENT (Privileged Communication)
Principal Investigator: John Snow
Review Group: Epidemiology
Meeting Date: March 4, 1853
Requested Start: August 1, 1853
Project Title: An Investigation into the Mode of Communication of Cholera
DESCRIPTION (provided by applicant):
The cholera has wreaked intermittent misfortune and death upon large swaths of civilization. While its reach extends ever farther, engulfing new port cities and populations, we have come no closer to fathoming the mode of its communication nor to stopping this awful malady. The disease affects the alimentary tract first and foremost, which implies strongly that one should look to contaminated water or food to explain its transmission. Its epidemic spread is along the pathways of human commerce, but it spreads no faster than people travel. As it usually appears first at seaports, it would appear to be spread by mariners, but it only affects mariners sailing from cholera-affected ports. There are numerous examples of the cholera apparently being transmitted by consumption of water polluted by excreta. Nonetheless, the hypothesis that cholera travels through humans and especially through contaminated water has not been put to crucial scientific examination. Herein, the PI proposes to conduct that crucial study.
Specifically, the PI shall collect data on cholera deaths during the next cholera epidemic to afflict London, focusing on deaths occurring in the area south of the Thames that is supplied by both the Lambeth and the Southwark and Vauxhall companies. The former company pipes comparatively clean upstream water to its users and the latter conveys water drawn from downstream, thereby being contaminated with the sewage from much of London.
These circumstances offer what the PI deems to be a natural experiment, as the population of this district of London has been divided into two groups consuming very different water, but the groups are otherwise indistinguishable from one another. The inhabitants seldom know the source of their water, and neighboring homes may subscribe to different water companies according to whatever deal the landlord struck. The consequence is that either clean or filthy water courses directly to hundreds of thousands of people living among one another, both rich and poor, of every rank and station, in large families and small, of all occupations and pursuits, living shoulder to jowl. To conduct my natural experiment, all that is needed is for me to learn the source of the water to the residence of each cholera victim, to estimate the denominators at risk for each of the two water sources, and then record the death rates according to the water source.
Thus, the research plan is for the PI to get the addresses of persons dying from cholera from the General Register Office, in those districts where the supply of the two Companies is intermingled in the manner described above. As it is sometimes difficult to obtain the receipts indicating the water source for a specific home, the PI proposes to use a secondary method to identify the water source for those who have died from cholera. Water from the two companies differs by a factor of 15 in its content of sodium chloride. Therefore, the PI will supplement his inquiry into the water source by using a chemical test that he has devised to ascertain the source of the water. The test involves adding nitrate of silver to a sample of water, which precipitates chloride from water, the results showing without question which company has supplied water to the household.
As the natural experiment has created two groups that appear to be alike in all relevant respects apart from whether their drinking water has been contaminated with sewage, the analysis will not be encumbered by the need to control for confounding factors. A simple table of deaths and people at risk among those consuming clean and contaminated water will suffice.
These are amalgamated remarks from a primary and two secondary reviewers.
Overall Impact: This application aims to evaluate the mode of communication of cholera, in relation to a specific theory about the quality of drinking water. Research that might lead to insight about slowing or stopping the spread of cholera merits high priority. The theory explored here, that cholera spreads through contaminated water, is one that carries little credibility in the scientific community. Nonetheless, we applaud the PI for pursuing this avenue of investigation despite its obvious limitations. The PI builds on his earlier arguments concerning evidence that cholera spreads through ingestion of water or food that has been contaminated with human waste. In supplementary material submitted with the application, the PI has suggested that the cholera poison must be able to reproduce itself within the body, and therefore must possess a structure that is most likely cellular, providing the apparatus for self-amplification. The grant proposal, however, does not acknowledge alternative, more plausible theories. Their absence must be considered an important drawback, lowering the priority for the proposal. We further believe that the proposal would have been stronger if the PI had more than one primary aim. For example, why not add as a secondary aim the possibility of spread through miasma? This more credible theory enjoys considerable support in the scientific community, and even if the PI does not assign it a high credibility, collecting evidence about miasmas would improve the overall priority of this application.
There were several other important limitations that influenced the overall assessment. We summarize these here and expand on them below: (1) the study population that has been proposed is not representative of the whole of London, much less of people outside of Britain who are afflicted with cholera; (2) the proposal is vague on how the denominators of the two water-company customers in the district of overlap will be measured; (3) misclassification of water sources is likely to be severe and to make the final results difficult to interpret; (4) the investigator, a general medical practitioner, has little experience in epidemiologic research, having spent much of his research time studying anesthesia in a laboratory setting, and there is little depth in terms of support staff; (5) no pilot data are available to indicate feasibility of the proposed approach; (6) the author gives little consideration to alternative and perhaps antecedent causes, such as poverty, crowding, and lack of general hygiene; and (7) the application indicates that there is little institutional support for the project.
The study addresses an important public-health problem that afflicts the population of Britain with increasing severity. If the study confirms the role of contaminated water in causing cholera (a big if), then it could conceivably lead to interventions that could mitigate the effect of cholera in the future.
As noted in the summary paragraph, the theory being investigated is not the leading theory about the communication of cholera. The consensus of the scientific community is that, while unsanitary conditions may contribute to cholera, the transmission of the disease is likely to be multicausal, a large component of which involves the action of miasmas that waft upward from cesspools and open sewers. Although the PI makes as strong a case as can be made for the role of water in transmitting the disease, it is not convincing. His preference for his water theory may be rendering him less objective about other possible explanations. Furthermore, his speculation that the cholera agent might be alive and reproduce within the alimentary tract is unsupported and incongruent with current biological knowledge, lowering the plausibility of his theory.
The investigator appears to be highly motivated and energetic.
Despite his enthusiasm for the project, there is little evidence to suggest that the PI has the credentials and experience for the kind of research he proposes. As documented by his CV, most of his previous research experience has been in the area of subjecting various animals to different doses of anesthetic gases. While there is no argument about the utility of those animal and bench studies, the investigator does not have a proven record of success in the kind of field work such as that proposed here. The work proposed is to be conducted largely by the PI himself, with hardly any research team to support the effort. We note, too, that the PI is engaged in a busy private practice as a general practitioner of medicine. No doubt his clinical experience has been useful in shaping his thinking on the topic, but we are skeptical that he will be able to find the time needed to carry through the work on the time scale that has been proposed.
The PI proposes to conduct a so-called natural experiment, the result of competition between water companies that has produced overlapping distributions of customers for the Lambeth and the Southwark and Vauxhall companies. This proposal has considerable merit, in that citizens consuming water from the upper Thames and the lower Thames are intermixed within some sectors of London. It seems useful to use this situation to test the theory about the water supply being the vehicle for transmission of cholera.
Unfortunately, the central idea for the study hinges upon the assumption that the denominator populations within the area of London with intermixed water supplies can be identified, or at least estimated. The proposal contains no information on how these denominators will be ascertained. It would seem to require a house-to-house survey in the area of overlapping supply, but no such survey, nor any alternative to it, has been proposed. To determine the source of water for the fatal cases, the PI proposes to conduct a chemical test that he devised himself, to distinguish, by virtue of its chloride content, the origin of the water within the household. Unfortunately, he has not yet published, nor does he include with the proposal any supporting information demonstrating the validity of his ad hoc chemical test, so we cannot evaluate how well this test would identify the water’s origin. The investigator does not seem to have considered that the salinity of the river will vary with the tides and the weather. We also emphasize that even if it were perfect, the chemical test would only identify the water exposure of the cases. The denominators would remain unknown. A second problem is the assumption that the water source to the home is sufficient to determine one’s exposure to Southwark and Vauxhall water. The PI has not even discussed the potential for misclassification that stems from water consumption during the workday or during evening or weekend visits. We believe that for this study to be carried out successfully, the investigator will need to conduct a survey with detailed interviews to determine customs for consuming drinking water. As the investigator’s theory does not specify how much of the supposedly contaminated water is needed to ignite the disease in a susceptible person, we cannot assess whether misclassification of the source of drinking water would interfere with a reasonable evaluation of the theory. What, also, of other beverages consumed? There is no intent to measure their consumption, nor tally the fraction of liquid intake that comes through the water companies’ pipes. Given these uncertainties, we are skeptical that the PI, even if he could collect the data he proposes to collect, would be able to evaluate his hypothesis with sufficient accuracy.
The design is novel and the proposal appears to include enough people to give robust results, although the PI should have included power calculations.
The study lacks a timeline, and given the lack of depth in research personnel, there is little assurance that it can be completed in a timely fashion. Furthermore, there are no pilot data to support the proposal, leaving it an open question whether the ambitious data collection proposed will be feasible. We believe that the PI has been overly optimistic in terms of requested budget and the anticipated work effort. The proposal lacks any detail on a statistical analytic plan. There is no mention of methods used to control confounding that may be identified after data collection. The PI seems to be naive in believing that the “natural experiment” will remove any confounding imbalances between his two cohorts of water consumers, and has not proposed any analytic methods to check for confounding or to address any confounding that might be detected.
The unique situation of citizens consuming water from sources different from their neighbors, via competing water companies in some districts of London, provides a great opportunity.
We believe that the proposal would have been stronger if the PI had forged some institutional ties, or had proposed to collaborate with the local sanitarian community and to integrate this project into a broader effort to study and control cholera. His lack of experience doing this type of research, his self-employment as a general practitioner of medicine with a practice largely devoted to administering anesthesia, his coolness for collaborating with the broader London medical community, and his single-minded attitude toward other currently debated scientific theories, all underscore our concern about the research environment and the likelihood that the PI can successfully conduct this work as proposed.
Regrettably, we cannot in good conscience recommend using scarce public-health monies to support this research. Any revised proposal would have to address the shortcomings above. Given the priority of the research topic, we encourage the PI to reconsider the proposal so that it could be competitive with other applications that deal with understanding and prevention of cholera.
ABOUT THE AUTHOR
KENNETH J. ROTHMAN is a Distinguished Fellow at the Research Triangle Institute, and Professor of Epidemiology and Medicine at Boston University. He is also the founding editor of EPIDEMIOLOGY. His research interests in epidemiology have spanned a range of health problems, including cancer, cardiovascular disease, neurologic disease, birth defects, injuries, environmental exposures, and drug epidemiology, but his main career focus has been the development and teaching of the concepts and methods of epidemiologic research.