Skip to content

The Times is out of joint. Sentinel testing in the foreign media.

9 May 2020

There was a small article in today’s paper, mentioning that the government had stopped paying the license fee for Pegasus software. Presumably, the AMLO administration, living up to its promises to stop spying on journalists and dissidents, no longer needs, or wants, the Israeli spyware, the Peña Nieto administration had bought for what was described in a New York Times article by Azam Ahmed back in 2017 as “an unprecedented effort to thwart the fight against the corruption infecting every limb of Mexican society.”   Putting the story into an international context, quoting experts (and victims), and ending with a quote by Carmen Aristegui (““For me, I have opted to believe that my public work is what will best protect me,” she said. “The great challenge for journalists and citizens is that the fear serve us, and not conquer us.”), Ahmad was right in submitting his piece to be considered for a Pulitzer Prize.

So, it’s not that Ahmad is … as some have it in the “cybersphere” over the last day… anything from an agent for imperialist dominance to a simple fool,  nor that he was, in a narrow sense, factually wrong, but that the story suffered from his choice of source material, and leads, once again, to serious questions about the Times‘ apparent bias when it comes to Mexican coverage (especially coverage of the 4th Transformation).

Ahmad’s main source was former Secretary of Health José Narro Robles, a surgeon turned politician with no background in epidemiology (and, perhaps, a less than stellar record in public health) during the Peña Neito administration, inheriting the dysfunctional IMSS public health system, which was as dysfunctional when he left office as when he entered.  Just with less money.  That the Times story was premised on the idea that the testing procedures used to track disease outbreaks in Mexico is woefully inadequate might have mentioned that Narro Robles never saw fit to update it, or question it until now.  Others quoted were working outside the country, none having access to the (unmentioned… see below) revisions to the so-called “Sentinel” system.  (In fairness, it might be possible that Ahmad’s story was submitted before Jorge Gallindo and Javier Lafuente’s study of the testing method was sent to El País (translated, with some simplification … honestly, to avoid having to deal with the charts and graphs, at the end of this post).

More seriously, like the Wall Street Journal‘s Juan Montes (“Death Certificates Point to Much Higher Coronavirus Toll in Mexico“), the story is heavy on the under-counting common everywhere in the world, as if it were a specifically Mexican issue… this despite repeated explanations by Undersecretary of Health for Epidemiology Hugo López-Gatell that the numbers are not a “true picture”, nor are intended to be taken as such.  As it is, a quick look at the Johns Hopkins Coronovirus Resource Center’s “Critical Trumps” mortality index shows Mexico as having about 76,000 deaths from COVID-19, to Mexico’s 3000. With Mexico a third the population of the United States, the country would have had to under-report by 65% to even match the REPORTED mortality in the US (which, as the New York Times itself reports, is woefully under-counted).

None of this probably would have mattered, given the general sense that the foreign media almost never puts a Mexican story into context, but this is the freaking New York Times… the most quoted foreign media source in Mexico (any Times story about Mexico will get a mention in the Mexican media).  It feeds into the sense on the political left that the corporate media, especially US corporate media, and The New York Times in particular, will always present the Morena government and the 4th Transformation in the worst possible light.  A feeling that’s hard to deny when the Times regularly opens its opinion columns to admitted political foes of AMLO, Enrique and León Krauze and Jorge Casteñada.  And, whether he is or isn’t biased, Ahmad’s reporting from Mexico in which he has relied on Carlos Loret de Mola (a good reporter, but with his subsidy from the former governments abruptly cut off, one who has become a mouthpiece for anti-AMLO politicians) and written flattering stories about Claudio X. Gonzales (the one-man “think tank” whose “anti-corruption” campaigns are designed to push a neoliberal privatization push).  It’s going to be a controversial topic, and the reporter’s motives, whether fair or foul, will be discussed.

Ahmad may have missed the point of what Carmen Aristigui said in his earlier piece:  ” “The great challenge for journalists and citizens is that the fear serve us, and not conquer us.”  A report that does nothing but raise fears… a far cry from the excellent critical analysis of testing and its limitations, from El País:


As with almost every country in the world, the real size of the SARS-CoV-2 contagion has haunted the Mexican government since the beginning of the pandemic, although questions have intensified in recent days. As the country enters phase 3, the issue has become critical. Many specialists complain of a lack of transparency and the absence of some data that, if made public, would provide more clarity. The Government, however, contends that its epidemiological surveillance model has been useful up to now to make mitigation decisions and has opted for another method to have greater control over prioritizing hospital needs in this new phase.

Since the beginning, the Government has accepted that the number of cases confirmed by diagnostic tests only represented a part of the epidemic. Until a few weeks ago, it offered its own estimate of the true dimension, based on the previous administration’s “Sentinel” epidemiological surveillance system. However, in recent days, the undersecretary of Health, Hugo López-Gatell, has been belittling this estimate, implying (without directly affirming it) that the degree of diffusion of the virus makes it, the estimate only a secondary tool for estimating actual cases. López-Gatell has emphasized that the Government’s efforts during phase 3 should focus on preventing hospitals from collapsing, avoiding situations in other world capitals like Madrid or New York… something Mexico has not experienced.

The number of diagnostic tests, used to gauge the size of the epidemic has never been a priority, but many citizens, and those in the scientific community, still want an estimate of the magnitude of the outbreak.

There are several aspects that inevitably contribute a certain amount of uncertainty and confusion about the future of the pandemic in the Latin American country. On the one hand, the lack of evidence: when it comes to testing, Mexico is at the tail of the Latin American countries. The highly praised Sentinel system used up to now has given way to a mathematical model, developed by the National Council for Science and Technology (Conacyt), although no further information has been provided so far on the latter. In addition, López-Gatell has admitted this week that there may be an underestimation of the death toll. If the data on mortality by causes for 2020 were made public, it would be feasible to identify those who died from symptoms similar to covid-19 and compare it with the average mortality rate from previous years.

Thursday (7 May) the Government reported 2,961 registered deaths and 29,616 reported cases. López-Gatell, during hsi daily progress report to the press, expressed the hope that herd immunity will be achieved, a phenomenon “in which a fundamental proportion of people necessary for the efficiency of contagions to no longer allow an epidemic to occur and it is controlled and disappears. We have a large number of cases, with mild symptoms, that we have asked to stay home. How many? Many, many. Hundreds of thousands. I wish they were millions, because that is what is going to stop the epidemic.” EL PAÍS estimates that Mexico would have between 620,000 and 730,000 symptomatic cases accumulated since the start of the pandemic, the majority being mild cases, for this to occur.

Evidence: problem #1

The most direct way of detecting the presence of a pathogen in a population is by testing suspected cases of infection. Although all the countries of the world have been struggling to expand their diagnostic capacity, some have more easily overcome them: in number of tests, Mexico is at the bottom of the continent, along with Brazil and Bolivia.

José Luis Alomía, general director of Epidemiology of the Ministry of Health, ensures that “so far more than 110,000 tests have been carried out.” In addition, he adds, “we have already paid the supplier the amount of 300,000 tests, which we hope to receive over the course of the weekend.” However, no change in testing strategy is noted that points to a significant change in the current trend.

With the incomplete perspective offered by the tests, Mexico can no longer give an exact number of infections: the value of 26,000 confirmed cases can hardly be considered as approximate. However, its health system has other instruments to expand and complement this number. One of them has been used from the beginning: the Sentinel epidemiological surveillance system.

Sentinel was installed and perfected during the past decade and a half, the H1N1 epidemic being a major spur to its development. The principle is simple: 475 of the 26,000 health care units in the country report weekly to the Ministry of Health weekly the frequency of consultations for Acute Respiratory Infection and Influenza Type Illness (ETI/SARI, for their initials in Spanish). Because, as López-Gatell himself stated, the symptoms of COVID-19 are indistinguishable from the ETI / IRAG, the existing “Sentinel” monitoring has been used as a first-pass diagnostic test.

As explained by Lopéz-Gatell on 3 May, a “double filter” is applied to the whole number of respiratory infections reported by the 26,000 health workers collecting the data. The percentage of “normal” ETI/IRAG infections. based on previous data is backed out, assumed to be “normal” ETI/IRAG cases. Of the 7.9 million consultations for respiratory causes identified though epidemiological week number 15 (April 5-11), the Ministry of Health considers that 899,374 or 5.9% would correspond to ETI / SARI. A sound filter of the remaining cases is a test for “positive” CORVID-19 infections. However, this data depends on reports by various agencies, collected in different data sets (by age, locality, etc.). Based on these reports there were 106,380 estimated cases of COVID-19 as of April 11, 2020.

This was the number released to the public at the insistence of Dulce Soto, a Reforma reporter. Although more recent data is available, the Ministry of Health has continued to base it’s “Sentinel” estimates on the 11 April “Sentinel” model as it enters “Phase 3” of the pandemic. Dr. Alomia says the old data is being used as “background”, “to project where in the country epidemic peaks will occur”, in order to plan for the availability of health service”. The Sentinel model was not intended for this, and although the 475 test sites continue to provide updated information, it is only a tool for deciding where and when to allocate resources.

To do this, Alomía explains that they have a different “mathematical model” developed by epidemiologist and mathematicians at the National Council for Technology and Science (CONACYT, for its Spanish acronym). According to the Health Ministry, this permits the government to know the approximate date when the epidemic will peak in a given locale — Cancun, Tijuana, Villahermosa or Mexico City, for example — when the demand for hospitalization will increase and of use in mobilizing resources, such as ventilators, monitors, specialized personnel to the areas where we are waiting for those peaks to occur and the beds are guaranteed. At this point, according to Alomía, “the simple fact of being able to estimate a total burden at the national level loses its usefulness, because it does not help us with all this. Now we no longer estimate based on what happened but on projections for the future”.

Asked the particulars of the model, the director of Epidemiology only replied that “it is the intellectual property of CONACYT” and, for now, it is not public. In the absence of a public version of the new territorial predictive model, and the apparent discontinuation of sentinel-based estimates, criticism of the Mexican government has increased. While much of the public believes they have been misinformed on the size of the contagion for the last three weeks, authorities have decided not to share (at least, as of yet) the information that are uses to navigate what is expect to be the worst period of the pandemic.


What is known about the new model is that it assumes those people tested by the Sentinel reporters who exhibit any two out of three symptoms (cough, fever and headache) rather than all three (which was reported as “suspected COVID-19) are assumed to be “suspected cases” and are added to the reported totals. It has been suggested that it might be prudent to consider persons with only one of three symptoms as a suspected case. This would make the Sentinel system work as it was intended, giving the data needed to plan for outbreaks of respiratory illnesses, assuming COVID-19 will be the most likely in some areas. Compared to reports from previous years, under the new criteria, the Ministry of Health estimates 43,000 more cases.

Although the 475 reporting sites are less than random in their sampling, it is estimated that it reflects 5.9% of all respiratory illnesses. From the data, Lopéz-Gatell reported an estimated 899,374 ETI / SARI infections in Mexico. That same figure for the current week is probably substantially higher.

Assuming that said projection by the Secretariat is accurate, and extrapolating this percentage, those 43,000 cases could mean up to 730,000 more cases in ETI / IRAG in 2020 compared to past years in the country.

It should be clarified that these estimated 730,000 cases are not directly translatable to COVID-19 without definitive verification by means of diagnostic tests, but if we assume that there is no other plausible cause to explain this increase in addition to the new virus, it would give a rough idea of the maximum infection rate by all respiratory infections, not just COVID 19.

No comments yet

Leave a reply, but please stick to the topic

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: