No announcement yet.

Case-control study to assess potential risk factors related to human illness caused by novel coronavirus (WHO, May 17 2013)

  • Filter
  • Time
  • Show
Clear All
new posts

  • Case-control study to assess potential risk factors related to human illness caused by novel coronavirus (WHO, May 17 2013)

    [Source: World Health Organization, full PDF document: (LINK). Edited.]

    Case-control study to assess potential risk factors related to human illness caused by novel coronavirus

    Date: 17 May 2013

    Contact: Dr Anthony Mounts,


    This document was adapted from a protocol developed by the Consortium for the Standardization for Influenza Seroepidemiology (CONSISE), a global partnership aiming to develop influenza investigation protocols and standardize seroepidemiology to inform public health policy for pandemic, zoonotic and seasonal influenza. This international partnership was created out of a need, identified during the 2009 H1N1 pandemic, for better (standardized, validated) seroepidemiological data to estimate infection attack rates and severity of the pandemic virus and to inform policy decisions. More information on the CONSISE network can be found on their website:


    This investigation will provide data to evaluate exposures and risk factors for human cases of novel coronavirus infection to determine those that are related to infection. This protocol outlines a case-control study and the epidemiological methods to guide data collection to assess risk factors for illness caused by nCoV infection. Health care personnel and the evaluation of other contacts are addressed in a separate protocol.

    Comments for the user?s consideration are provided in purple text throughout the document as the user may need to modify methods slightly because of the local context in which this study will be carried out.



    The novel coronavirus, sometimes referred to as the Middle Eastern Respiratory Syndrome coronavirus (MERS-CoV) was first detected in a patient living in Saudi Arabia in September of 2012. Since that time, it has returned in both sporadic cases, in small clusters, and in large outbreaks.

    While the source of the virus is currently unknown, it is thought to originate in animals. Human-to-human transmission has also been documented on multiple occasions. Although finding the putative animal reservoir is an important step in controlling spread of the virus, a more immediate need is to understand the route and mode of transmission to humans, and the types of exposures that result in infection. Several possibilities exist, including direct contact with an infected animal, which could be either the reservoir species or an intermediate host species; contact with or consumption of unprocessed animal products; contact with the environment where an infected animal has recently been; or consumption of a food or beverage which has been contaminated by animal excreta. All of these have been implicated in other zoonotic infections. Learning the mode of transmission to humans will allow measures to be taken to interrupt transmission. This investigation will provide data to evaluate risk factors for infection by reviewing exposures of known cases and comparing them to rates of exposure in similar uninfected individuals in the general population.

    Current information on the novel coronavirus and cases can be found on the WHO website:

    A separate protocol is available specifically for assessment of exposed Health Care Personnel and contacts of cases (see ).


    The data collected from this study will be used to refine/update recommendations for surveillance and case definitions, to characterize the key epidemiological transmission features of nCoV virus, help understand spread, severity, spectrum of disease, impact on the community and to inform operational models for implementation of countermeasures such as case isolation, contact tracing and quarantine.
    The primary objective of this study is to:
    • Identify modifiable non-human exposures that lead to human nCoV infection.



    The study uses a case-control design that examines the differences in types of exposures between symptomatic individuals with laboratory confirmed nCoV infection and healthy controls in order to determine the risk associated with that exposure. A standard questionnaire has been provided in this protocol (appendix A), however, the study must begin with less structured interviews of confirmed cases (or their proxies if unable to respond due to critical illness or death) in order to develop hypotheses about the possible exposures. Once these hypotheses have been developed, they can be tested through this case control study. For example, many of the cases might report that they ate a fruit such as dates in the week before they became ill. However, many people in the region eat dates frequently. The purpose of the case control study is to determine if a case was more likely to have eaten dates in the 10 days before he became ill than a similar uninfected person in the community.

    Ideally, this study should be done prospectively, enrolling new cases as they are identified and selection controls at the same time. However, it will be necessary to do the study retrospectively for many of the cases as they occurred some time ago. For retrospective cases the period of interest is still the 10 days before the onset of their illness however as the controls have not been sick, either the same 10 day period can be used if the elapsed time is short enough to allow sufficient recall or the 10 day period before the questionnaire is administered. Controls should be chosen at random from the area of residence of the cases and of the same age and sex as the cases.


    Study subjects will include only cases who are known to have been infected with nCoV through laboratory confirmation (see section 2.4.2 below). If the case is part of a cluster, only the case which is thought to be the ?index case? or first person infected in the cluster will be included.

    This will exclude cases for whom the transmission may have occurred from exposure to the index case (i.e. human-to-human transmission). While there are many questions about the mode of transmission and risk factors for human-to-human transmission, the primary purpose of this study is to determine the non-human source of infection, such as exposures to animals or contaminated food products. Probable cases should be excluded from this study because the current probable case definition requires exposure to a known nCoV case in order to be classified as probable.


    Case definitions for reporting are provided by WHO and are subject to change as more information becomes available.1


    A confirmed case of nCoV is a person with laboratory confirmation of infection with nCoV, either through direct detection of the virus (by viral isolation) or viral RNA (by molecular methods such as RT-PCR) in clinical specimens.

    Currently, serological testing has not been used as a confirmatory test as assays are still in development. However, in the future the confirmed case definition may change. Investigators are encouraged to refer to the WHO website for current recommendations and case definitions.




    Primary study subjects are laboratory confirmed cases of nCoV infection. If cases are part of a cluster, only the index case ? that is, the case with the earliest date of onset of illness ? should be included in the analysis. (note, it may be useful to collect the same data from all of the cases in the cluster as there are a number of other questions that may be studied but for the purposes of this study, only the index case will be included.)
    As stated above, probable cases should generally not be included in the analysis as they by definition are cases with exposure to known cases and as such are part of a cluster in which nCoV transmission may have occurred from person to person. The one exception may be if during an interview with a confirmed case, it is discovered that a contact, such as another family member, would meet the definition for probable case AND that the probable case occurred earlier, making him or her the likely index case.

    COMMENT: In some clusters, more than one index case may appear simultaneously (co-primary/co-index cases) and it is possible to tease out the transmission dynamics and identify the index case; another option for preliminary analysis is to exclude these clusters.


    To understand how rates of exposures to potential sources of infection differ between cases and uninfected individuals, it is necessary to recruit age and sex matched control subjects who live in approximately the same neighborhood as the case. Controls should be randomly selected as described below and asked for their consent to participate in the study.

    To maximize the power to show differences in exposures, it may be desirable to recruit as many as four controls for each case.

    Controls should be of the same sex as their respective case and from the same general area where the case lives.

    Age matching can be done within a range that depends on the age of the case for which the control is being selected, as outlined below:

    [Age of case - Age range of control]
    • < 5 y.o.- Within ?1 year of age
    • 5 to 18 y.o. - Within ? 3 years of age
    • ≥18 y.o. - Within ? 5 years of age
    COMMENT: Currently, circulation of this virus in the community is thought to be nonexistent or minimal at most and the numbers of infections low. For that reason, prospective controls who have not had recent respiratory illness can be enrolled without laboratory confirmation of the absence of nCoV infection. However, if the situation changes in the future, it will be necessary to exclude controls from the study who are determined to have had nCoV infection through serological testing.


    Controls will be selected randomly. There are several ways to do this, the simplest is the most direct ? to go to the area where the case lives and select them directly through a random selection process on site. Once arriving at the home of the case, select the closest dwelling to the case home. If a decision is needed about the closest dwelling, for example if there are two dwellings on either side equal distance from the case home, decide the direction by a coin toss. When arriving at the home, inquire about the members of the household to determine if any residents would meet the requirements for a control in terms of matching the case in age and sex. If multiple members of the household would qualify, choose by a coin toss (if only two) or drawing numbers at random. Only use one member of a given household for a control. Continue on until four controls are selected for each case. If the residence is a multi-family dwelling such as an apartment building, choose the floor to start on by random number drawing. Choose the direction to proceed on the selected floor by a coin toss. Choose the first apartment on the floor, if there are multiple, by random number drawing. If no suitable control is found in the first apartment, continue through consecutive apartments on the floor after choosing the direction from the first apartment by coin toss. Alternatively, a simple random sample of families in the neighborhood can be selected using random numbers and a list of households, if available.


    During the visit to both cases and controls, the purpose of the study will be explained to all eligible subjects and their consent obtained by a trained member of the investigation team. Consent for children under the age of 18 years old will be obtained from their parents or guardians. Verbal assent will also be obtained for children under 17 years old.

    COMMENT: The age of consent may vary by country. Check with local IRB requirements.


    After enrollment and informed consent is obtained, a standardized minimum data set will be collected. A template of the study questionnaire for the use of all cases and contacts can be found in Appendix A. These include some identifying information, demographic information, and date of onset.


    In addition to the minimum dataset, detailed questions to evaluate risk factors for human infection with nCoV will be asked in a questionnaire (Appendix A). These are included in the data collection form in under the section for ?exposures?. These questions are more specific and include aspects of timing of, frequency and duration of exposure(s).


    Prior to study implementation, front-line staff including all study personnel will be trained in infection control procedures (standard, contact, droplet or airborne precautions) including proper hand hygiene and the correct use of surgical or respiratory face masks, if necessary, not only to minimize their own risk of infection when in close contact with patients during home visits and elsewhere, but also to minimize the risk of the personnel acting as a vector of nCoV transmission between subjects members or between households. If N-95 respirators are to be used, they should be fit tested in advance.



    Laboratory testing is necessary to prospectively identify cases and, if needed, to eliminate individuals who have been infected from the control group. Please refer to the current WHO laboratory guidance for appropriate collection, storage, shipment, and testing procedures. Current WHO lab guidance is available at:

    Additional records should be kept for each biological sample, including the time of collection, the conditions for transportation and the time of arrival at the study laboratory.



    COMMENT: Limited available data suggests that lower respiratory specimens such as sputum, endotracheal aspirates, or brochoaveolar lavage, will have a higher yield for testing than upper respiratory specimens such as nasopharyngeal swabs. Ideally, multiple respiratory specimens should be collected from different respiratory sites on multiple days to maximize detection of nCoV. A person suspected to have nCoV infection but who has a nasopharyngeal swab collected that tested negative for nCoV should have repeat testing on subsequent days if strongly suspected to have nCoV infection. nCoV has also been found in stool of infected patients, which may provide an additional of confirmation.

    To consider a case as laboratory-confirmed, one of the following conditions must be met:
    • positive RT-PCR or other validated molecular assays for at least two different specific targets on the nCoV genome
      • OR
    • one positive RT-PCR assay for a specific target on the nCoV genome and an additional different PCR product sequenced, confirming identity to known sequences of the new virus.
    A positive RT-PCR assay for a single specific target without further testing is considered presumptive evidence of nCoV infection. Final classification of cases will depend on clinical and epidemiological information combined with laboratory data. Case definitions can be found at:

    Member States are requested to immediately notify WHO.

    See full details for virologic laboratory testing of nCoV can be found here:

    COMMENT: there is currently no provision for case confirmation using a serological assay, however, this may become available in the near future. Investigators should refer to current WHO case definitions for determining case classification.


    The methods described in this section may be useful if it becomes necessary to exclude infected individuals from the control group.
    COMMENT: Validated serologic assays for nCoV are currently limited but are being pursued by a small number of laboratories across the globe. Here we provide details of the only published serologic testing available for nCoV.1,2

    COMMENT: Only a limited number of laboratories have the facilities for nCoV serologic testing and therefore collaboration between countries without current capacity and designated reference laboratories is possible.

    Collaboration is up to the discretion of member states carrying out the research, but WHO/EMRO strongly support such collaboration and would willingly facilitate collaboration and possible shipment elsewhere for testing.

    The following laboratory assay results are currently available for defining a case as nCoV antibody positive and full details can be found in1,2.
    • Screening for antibodies reactive to nCoV by indirect immunofluoresence assay (IFA) described by 1,2
    • It is strongly recommended that confirmatory serologic testing should be done using microneutralization or ELISA-based assays using appropriately timed sera (ideally paired acute and convalescent sera)1,2

    • Ethical approval will be sought in accordance with local, regional and national authorities.
    • Comment: It is strongly recommended that ethical approval is obtained in advance from relevant ethical or institutional review boards (e.g., national Ministries of Health, Agriculture, etc.) using a generic protocol such as this one prior to an outbreak. Once an outbreak occurs, the study design, questionnaires, sampling and consent forms can be modified rapidly to the actual situation. This may still have to be resubmitted for ethical approval, but as the generic protocol including this final step has already been approved, this could be a very rapid process, without substantial delay to the start of the investigations.


    The following section discusses the endpoints ? that is, what will be measured and calculated using the data that are collected in this study ? for the primary objectives, including statistical advice.



    The following will be assessed as study endpoint corresponding to the study?s primary objective:
    • The ratio of the odds of exposure vs. odds of no exposure to a variety of potential sources of infection in cases vs. controls.
    • The exposure or combination of exposures that best explain the resulting infection based on a regression model of all exposures.




    The reported practices among cases and matched controls should be compared using appropriate statistical tests, e.g., Bivariate associations between risk factors and infection will be determined by chi-square statistics or 2-sided Fisher?s exact test and expressed as odds ratios with 95% confidence intervals. Multivariable logistic regression will be used to further analyze the associations to determine which best explains the resulting infection.

    COMMENT: Univariate statistical analysis by use of logistic regression for a case‐control study could be used to test the significance of each predictor on the outcome of infection. Multivariable logistic regression can be used to identify a combination of risk factors associated with the odds of infection.

    COMMENT: Alternatively, Mantel-Haenszel matched-pair analysis (McNemar test) can be used to estimate the strength and statistical significance of associations between exposures and infection.3


    Any deviations of the study methodologies should be reported along with the results to aid in the interpretation of findings and to assist others in improving future versions of the protocol.


    Although not described as part of this investigation, we recommended that in conjunction with this case control study , full epidemiological and virological outbreak investigations around new cases, including (1) further investigations of close familial, social and health care worker contacts, (2) environmental sampling including testing of areas around the infected household, farms, markets and potential contaminated water sources and (3) retrospective animal mortality investigations should supplement these activities in collaboration with relevant parties, in particular if the objective would include identifying a zoonotic source of infection among index and/or contacts of the index.


    Papers related to nCoV:Risk factor investigations of other novel viruses:
    • Dinh P, Long H, Tien N, et al. Risk factors for human infection with avian influenza A H5N1, Vietnam, 2004. Emerg Infect Dis 2006;12(12):1841-7.
    • Reynolds MG, Anh BH, Thu VH, et al. Factors associated with nosocomial SARS-CoV transmission among healthcare workers in Hanoi, Vietnam, 2003. BMC Public Health 2006;6(207):207 doi:10.1186/1471-2458-6-207.
    • Zhou L, Liao Q, Dong L, et al. Risk Factors for Human Illness with Avian Influenza
    • A (H5N1) Virus Infection in China. JID 2009:199 (15 June): 1726-34.
    • Areechokchai D, Jiraphongsa C, Laosiritaworn Y, et al. Investigation of Avian Influenza (H5N1) Outbreak in Humans ? Thailand, 2004. MMWR vol. 5 suppl. April 28 2006.
    • Mounts A, Kwong H, Izurieta HS, et al. Case-Control Study of Risk Factors for Avian Influenza A (H5N1) Disease, Hong Kong, 1997. JID 1999;180:505-8.


    Questionnaire for case-control study to assess risk factors related to human illness caused by novel coronavirus

    The following questionnaire should be used for all cases and controls included in the investigation. The time frame for the questions to cases is the 10 day period before the onset of their illness. For cases that happened in the past, it is useful to use memory prompts such as birthdays, holidays, or other memorable events that occurred around the same time to help the interviewee recall the specific time frame of interest. For controls, ideally the same time frame would be asked about as for the cases, however for retrospective cases in the remote past, this is not generally feasible. In that situation, the 10 day period just before the interview can be used as a reference. For cases that are enrolled prospectively when they are diagnosed, controls should be chosen at that time and the time frame of reference for the questions should be the same as for the case.

    If the case has died or is otherwise unable to answer questions, a proxy such as a close relative who knows the person well can answer the questions for him or her.

    Each subject should be allocated a unique identification number.

    COMMENT: Once questionnaire is finalized, full instructions and skip patterns should be added. Comments throughout the questionnaire are highlighted with purple text.

    COMMENT: Note that adding multiple choice answers will allow for easier data analysis.



    (1) WHO Case Definition for nCoV: