BMJ. Detection of mild to moderate influenza A/H7N9 infection by China?s national sentinel surveillance system for influenza-like illness: case series

[Giuseppe]

[Source: British Medical Journal, full page: (LINK). Abstract, edited. h / t CIDRAP.]

Research

Detection of mild to moderate influenza A/H7N9 infection by China?s national sentinel surveillance system for influenza-like illness: case series

<CITE><ABBR>BMJ </ABBR>2013; 346 - doi: http://dx.doi.org/10.1136/bmj.f3693 (Published 24 June 2013)</CITE>
<CITE></CITE>
<CITE>Cite this as: <ABBR>BMJ</ABBR> 2013;346:f3693</CITE>
<CITE></CITE>
<CITE></CITE>
<CITE></CITE>Dennis KM Ip, clinical assistant professor 1, Qiaohong Liao, public health officer 2, Peng Wu, post doctorate fellow 1, Zhancheng Gao, professor and respiratory physician 3, Bin Cao, professor and infectious disease physician 4, Luzhao Feng, public health officer 2, Xiaoling Xu, respiratory physician 5, Hui Jiang, public health officer 2, Ming Li, public health officer 2, Jing Bao, respiratory physician 3, Jiandong Zheng, public health officer 2, Qian Zhang, public health officer 2, Zhaorui Chang, public health officer 2, Yu Li, public health officer 2, Jianxing Yu, public health officer 2, Fengfeng Liu, public health officer 2, Michael Y Ni, clinical assistant professor 1, Joseph T Wu, associate professor 1, Benjamin J Cowling, associate professor 1, Weizhong Yang, medical epidemiologist and deputy director 6, Gabriel M Leung, professor 1, Hongjie Yu, medical epidemiologist and director2

Author Affiliations: <SUP>1</SUP>Infectious Disease Epidemiology Group, School of Public Health, Li Ka Shing Faculty of Medicine, University of Hong Kong, Hong Kong Special Administrative Region, China <SUP>2</SUP>Division of Infectious Disease, Key Laboratory of Surveillance and Early-warning on Infectious Disease, Chinese Center for Disease Control and Prevention, Beijing 102206, China <SUP>3</SUP>Department of Respiratory and Critical Care Medicine, Peking University People?s Hospital, Beijing, China <SUP>4</SUP>Department of Infectious Diseases and Clinical Microbiology, Beijing Chao-Yang Hospital, Beijing Institute of Respiratory Medicine, Capital Medical University, Beijing, China <SUP>5</SUP>Department of Respiratory Medicine, Anhui Province Hospital, Hefei, China <SUP>6</SUP>Office of the Director, Chinese Center for Disease Control and Prevention, Beijing, China

Correspondence to: G M Leung gmleung@hku.hk and H Yu yuhj@chinacdc.cn

Accepted 5 June 2013


Abstract

Objective

To characterise the complete case series of influenza A/H7N9 infections as of 27 May 2013, detected by China?s national sentinel surveillance system for influenza-like illness.


Design

Case series.


Setting

Outpatient clinics and emergency departments of 554 sentinel hospitals across 31 provinces in mainland China.


Cases

Infected individuals were identified through cross-referencing people who had laboratory confirmed A/H7N9 infection with people detected by the sentinel surveillance system for influenza-like illness, where patients meeting the World Health Organization?s definition of influenza-like illness undergo weekly surveillance, and 10-15 nasopharyngeal swabs are collected each week from a subset of patients with influenza-like illness in each hospital for virological testing. We extracted relevant epidemiological data from public health investigations by the Centers for Disease Control and Prevention at the local, provincial, and national level; and clinical and laboratory data from chart review.


Main outcome measure

Epidemiological, clinical, and laboratory profiles of the case series.


Results

Of 130 people with laboratory confirmed A/H7N9 infection as of 27 May 2013, five (4%) were detected through the sentinel surveillance system for influenza-like illness. Mean age was 13 years (range 2-26), and none had any underlying medical conditions. Exposure history, geographical location, and timing of symptom onset of these five patients were otherwise similar to the general cohort of laboratory confirmed cases so far. Only two of the five patients needed hospitalisation, and all five had mild or moderate disease with an uneventful course of recovery.


Conclusion

Our findings support the existence of a ?clinical iceberg? phenomenon in influenza A/H7N9 infections, and reinforce the need for vigilance to the diverse presentation that can be associated with A/H7N9 infection. At the public health level, indirect evidence suggests a substantial proportion of mild disease in A/H7N9 infections.

-
-------

[gsgs]

Re: BMJ. Detection of mild to moderate influenza A/H7N9 infection by China?s national sentinel surveillance system for influenza-like illness: case series

> indirect evidence of a substantial proportion of mild disease
> It remains unknown whether the clinical iceberg phenomenon applies to A/H7N9
> Evidence suggests that there is an important proportion of mild disease, and supports
> the existence of a clinical iceberg phenomenon in influenza A/H7N9 infections

"5","substantial","important"

an earlier study concluded:

http://www.flutrackers.com/forum/showthread.php?t=205489

testing of samples from 20,739 patients who had ILIs revealed only six cases of H7N9 infection
We did not find evidence of widespread mild disease


depends all on how you formulate it

http://www.flutrackers.com/forum/showthread.php?t=205425&highlight=20739

5000 cases, 36 deaths , CFR=0.7&#37;
excluding those excess deaths from cardiovascular or respiratory
disease where influenza is not detected.
These may happen with H7N9 too. For seasonal flu this would be
an additional CFR of 0.1% , so my estimate CFR(H7N9)=0.8%

[Giuseppe]

Re: BMJ. Detection of mild to moderate influenza A/H7N9 infection by China?s national sentinel surveillance system for influenza-like illness: case series

The number of assumptions is skyrocketing as the number of possible cases. And this is not much informative.

554 sentinel points
12.5 swabs per week / point
554*12.5*4 wk=27,700 swabs per month

5 sentinel swabs tested positive for H7N9 from the above paper =

5 / 27,700 = 0.02% positivity rate

If in a 5,000,000 China megacity during the same period of time there was an ILI rate of 14 x 1,000 then:

14 : 1,000 = x : 5,000,000

x = 70,000 ILI cases per week

if these ILI cases should have been swabbed - employing the above positivity rate we should have had:

70,000 * 0.02% = 14 H7N9 cases

[NS1]

Re: BMJ. Detection of mild to moderate influenza A/H7N9 infection by China?s national sentinel surveillance system for influenza-like illness: case series

Scalability Questions
Around
Rate of Mild Case Detection

The number of assumptions is skyrocketing as the number of possible cases. And this is not much informative.

554 sentinel points
12.5 swabs per week / point
554*12.5*4 wk=27,700 swabs per month

5 sentinel swabs tested positive for H7N9 from the above paper =

5 / 27,700 = 0.02&#37; positivity rate

If in a 5,000,000 China megacity during the same period of time there was an ILI rate of 14 x 1,000 then:

14 : 1,000 = x : 5,000,000

x = 70,000 ILI cases per week

if these ILI cases should have been swabbed - employing the above positivity rate we should have had:

70,000 * 0.02% = 14 H7N9 cases

Scaling

Scaling statistics using national factors onto discrete (sub-component) populations is not generally held as an ideal mechanism for creating actionable results at the discrete population level. When this type of spreading is undertaken at a giga-scale (i.e. population of China), often additional considerations and cautions must be employed. The types of projections are the purest form of invention.

An inspection of significant digits may be in order for these calculations while keeping at the forefront the fact that averages and assumptions are being multiplied by averages and assumptions. We will examine here the individual factors and the calculation veracity. Projection estimates will be provided at the end of this post for your example population of 5 million and for several actual geographies in China related to emergent H7N9.

You certainly are accurate on the skyrocketing effect of assumptions. One paper that you posted requires a minimum of 15 swabs per week and this paper suggests a quota range of 10 to 15 swabs per week, a ceiling 50% above the floor (substantial variance, σ sqared=6.25). Using actual counts from the earlier study (46,807 swabs from 554 locations over 8 weeks) gives an average of 10.56115 swabs per location per week.

Factor Validity

The authors openly admit that swabbing is not random and then they immediately attempt without any observational evidence to attach the idea that this quota-driven, non-random test is unbiased? They pledge that their system, though structured with a non-randomised collection, is representative of a billion count population? The system begins by missing 97.5% of the hospitals (sampling 2.5%) and then proceeds to require 15 swabs per week, but only gets about 10.56115 across a week's duration. We're not certain if the authors have supplied a supplement denoting the actual geographic collections? If an attempt is being made to differentiate disease epidemiology, then we'll need specific geographic counts rather than obtuse gerrymandering accumulations and averages.

Severity bias is generally present in infectious disease collections that are not intentionally randomised.

Test Duration

Is the test duration for this paper defined or are we to assume that the accumulations take place over the entire emergence period from their first publicly reported case in late February until May 27? If so, then 12 weeks very much dilutes the outcome when used in place of the 4 week estimate to accommodate the 5 mild cases discovered via passive surveillance of ILI reports. Early multipliers reduced by 66.67% dampen results.

Emerg Infect Dis. Monitoring Avian Influenza A(H7N9) Virus through National Influenza-like Illness Surveillance, China [Thread #205425, Post#1] (emphasis mine)

On April 3, 2013, to enhance surveillance for influenza A(H7N9) virus, all network laboratories were required to increase the number of specimens to a minimum of 15/week and to test all specimens collected since March 4, 2013, for influenza A(H7N9) virus by real-time reverse transcription PCR . . .
. . .
During March 4–April 28, CNISN tested 46,807 nasopharyngeal swab samples from 554 sentinel hospitals throughout mainland China.

ReCalculation

Information Quality is very low in the emergent H7N9 discussion though Information Quantity sometimes exceeds necessity.

Using the actual numbers from the previous study (46,807 covering 8 weeks from early March to late April) summed with the mandated 15 swabs per week per hospital (33,240 covering 4 weeks from late April to late May), we estimate the denominator to be a minimum of 80,047 total swabs over 12 weeks. With detection of 5 mild cases, the rate calculates to:

Mild Case Detection Rate for Emergent H7N9 in ILI Cases ~ 0.006246329%.

Holding constant the ILI factor discussed of 0.01400 (14 Reports per 1,000 Population), the following selected geographic projections may be discussed based on the limitations of the incoming factors' veracity. Particularly daunting is the non-applicability of a national Mild Case Detection Rate across disparate, sub-state geographies.

Concentrated City
Population . . . . . 5,000,000
Projected ILI . . . . . .70,000
Projected Mild H7N9 Count___4.4

Beijing
Population . . . . 20,690,000
Projected ILI . . . . .289,660
Projected Mild H7N9 Count__18.1

Shanghai
Population . . . . 23,470,000
Projected ILI . . . . .328,580
Projected Mild H7N9 Count__20.5

Guizhou Province
Population . . . . 34,690,000
Projected ILI . . . . .485,660
Projected Mild H7N9 Count__30.3

Fujian Province
Population . . . . 37,200,000
Projected ILI . . . . .520,800
Projected Mild H7N9 Count__32.5

Zhejiang Province
Population . . . . 54,630,000
Projected ILI . . . . .764,820
Projected Mild H7N9 Count__47.8

Jiangsu Province
Population . . . . 78,990,000
Projected ILI . . . 1,105,860
Projected Mild H7N9 Count__69.1

China
Population . . 1,344,000,000
Projected ILI . . .18,816,000
Projected Mild H7N9 Count__1,175.3

As Chris & Edgar like to say, "Caveat Lector."

Genetic Calibration

We do tend to agree with the authors that cross-referencing the actual disease (gene form) to the statistics is potentially beneficial for the public.

Related Reading

• GeneWurx Cross Serotype Homology Analysis, Open-Access, Full-Text version
• Human Emergent H7N9 from Zhejiang Province
• Environment Emergent H7N9 from Zhejiang
• Fatal H5N1 Homology to Emergent H7N9 from Shanghai in March
• Mammalian Adaptation of Emergent H7N9 Neuraminidase in Jiangsu
• TamiFlu Resistance via NA 292K in Taiwan and Shanghai
• Zhejiang Fatal H7N9 with Divergent NA, PB2 701N and Aberrant PA
• Novel PB2, Divergent PB1 & Novel PA from First Fujian Case
• China - H7N9 Human Isolates on Deposit at GISAID, Comprehensive Genetics Discussion

[sharon sanders]

Re: BMJ. Detection of mild to moderate influenza A/H7N9 infection by China?s national sentinel surveillance system for influenza-like illness: case series

I think the only conclusion we can draw from this paper is that healthy children, aged 13 and younger, generally have good outcome after infection with H7N9.

We know this from the confirmed infections referenced on our case list. Those who were hospitalized were in stable or good condition. At least one was confirmed retroactively and recovered with no treatment:


#14 - Child, 4, onset date March 31, mild case. Shanghai

#44 - Child, 7, hospitalized April 11 Shunyi District Beijing

#61 - Child, 4, PCR positive in routine screening of contacts/neighbors of case #44 - asymptomatic Beijing

#76 - Child, 2, Onset March 17 - recovered, Shanghai treated in Hunan Province Note.

#127 - Child, 4, onset date April 27, Zaozhuang City in Central City, Shandong Province

#131 - Child (male), 9, Onset on April 26. Currently recovered and discharged after mild illness. Cangshan District, Fuzhou. Fujian province

#133 - Boy, 6, onset date May 21, no reported contact with poultry, condition good - Beijing


-----------------------

It appears from the small sample size of 133 confirmed cases, that young children have a better outcome than senior citizens and that it is possible that there are many, many cases of asymptomatic and/or mild H7N9 influenza virus among young children that are not quantified.

[NS1]

Re: BMJ. Detection of mild to moderate influenza A/H7N9 infection by China?s national sentinel surveillance system for influenza-like illness: case series

I think the only conclusion we can draw from this paper is that healthy children, aged 13 and younger, generally have good outcome after infection with H7N9.

We know this from the confirmed infections referenced on our case list. Those who were hospitalized were in stable or good condition. At least one was confirmed retroactively and recovered with no treatment:


#14 - Child, 4, onset date March 31, mild case. Shanghai

#44 - Child, 7, hospitalized April 11 Shunyi District Beijing

#61 - Child, 4, PCR positive in routine screening of contacts/neighbors of case #44 - asymptomatic Beijing

#76 - Child, 2, Onset March 17 - recovered, Shanghai treated in Hunan Province Note.

#127 - Child, 4, onset date April 27, Zaozhuang City in Central City, Shandong Province

#131 - Child (male), 9, Onset on April 26. Currently recovered and discharged after mild illness. Cangshan District, Fuzhou. Fujian province

#133 - Boy, 6, onset date May 21, no reported contact with poultry, condition good - Beijing


-----------------------

It appears from the small sample size of 133 confirmed cases, that young children have a better outcome than senior citizens and that it is possible that there are many, many cases of asymptomatic and/or mild H7N9 influenza virus among young children that are not quantified.

Sharon, you are right on track with this observation. Perhaps the researchers will consider a series of focused serology studies based on the evidence you've presented.

As we've been discussing, their current reliance on the shotgun approach in a disparately distributed population produces a certain quantity of information, but that information is doomed to low quality due to skewing and bias. Selecting targeted populations and intensively collecting discrete and well-defined data points allow modeling that is much closer to representing reality and much more efficient for discovering causal factors.

You've now provided a foundation that any research team can use to produce valid working hypotheses "FOR" and / or "AGAINST" widespread testable, but sub-clinical emergent H7N9 infection in the most important segment of the human population based on YLL projections. Will one team say, "Yes," to the project?

We'd all like to know who is at risk in a pandemic?