Flu Effects on US Health Demand

Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0

Richard Evans / Scott Hinds / Ryan Baum


203.901.1631 /.1632 / .1627

https://twitter.com/images/resources/twitter-bird-blue-on-white.png richard@ / hinds@ / baum@sector-sovereign.com


January 16, 2013

Flu Effects on US Health Demand

  • We estimate that 4Q12 saw roughly 28 more flu-related hospital admissions per 100,000 population than 4Q11; this implies a +/- 1 percent flu-related gain in health services demand yoy
  • The ultimate duration and severity of the current season cannot be known with any certainty; however serology offers at least some indication. Predominant serology in the current season is type A (H3N2); past H3N2 predominant seasons have been longer, more severe, and produced more hospitalizations than average
  • Of the 16 H3N2 seasons for which we have hospitalization estimates, the average season saw an excess of 57 hospitalizations per 100,000 persons, with a range of 26 to 103. As of the first week of January, the current season already had produced cumulative hospitalizations / 100,000 of roughly 36, well above the low end of the historic H3N2 range
  • Assuming 1Q13 sees cumulative hospitalizations peak anywhere between 57 and 103 per 100,000 (mean and max for prior measured H3N2 seasons), we would expect 1Q13 flu-related per-capita demand effects of between +/- 43bp and +/- 2 percent, respectively

The ’12 /’13 influenza season is off to a relatively early and intense start; as of the first week of January 2013, patients with a serologically confirmed diagnosis of influenza have been admitted to hospitals at a cumulative rate of approximately 12 admissions per 100,000 population. Exhibit 1 compares hospital admission rates for confirmed-flu patients across all flu seasons since 2003

The hospitalizations tracked in Exhibit 1 are limited to cases in which influenza is confirmed; we are not immediately aware of any data that provide a ready basis for estimating total flu-related hospitalizations from this base of confirmed cases. As an extremely rough proxy, we note that the average number of hospitalizations (16.3 / 100,000 population) for cases of confirmed flu in the 9-year time series behind Exhibit 1 is roughly one-third the average rate (45 / 100,000 population) for estimates of total (confirmed or not) flu-related hospitalizations in the 22 flu seasons beginning in ’69/’70 and ending in ’00/’01[1]

Thus as a very crude, but still directionally useful estimate of flu-effects on hospital demand, we can estimate that the ’12/‘13 flu season has produced +/- 36 hospitalizations / 100,000 population (i.e. 3x the serologically confirmed rate) through the 1st week of January. Framing this in terms of quarterly growth, confirmed flu admits per 100,000 population stood at a cumulative 9.8 at the end of 4Q12, as compared to 0.4 at the end of 4Q11. Again assuming that total flu-related hospitalizations are roughly three times the level of confirmed cases, this would indicate that 4Q12 flu-related admits were 29.4 per 100,000 population, as compared to 1.2 in 4Q11. ‘Normal’ all-cause rates of hospitalization are roughly 2,940 admits / 100,000 population each calendar quarter, so we can very roughly estimate that flu increased the rate of hospitalizations by (29.4 – 1.2) / 2,940, or by +/- 1%, in 4Q12 as compared to 4Q11 (Exhibit 2)

Absolute flu effects in 1Q13 should be substantially larger than in 4Q12, both because the ’12/’13 season is off to a much earlier start than ’11/‘12, and because the current season has already surpassed the cumulative hospitalizations peak of the ’11/’12 season (Exhibit 1, again). However the relative (i.e. growth) effect of flu in 1Q13 is potentially reduced by the facts that: 1) a potentially large percentage of total ’12/’13 flu-related hospitalizations may have fallen into 4Q12; and, 2) hospitalizations associated with the ’11/’12 season were concentrated in 1Q12, making for a more difficult comparison

There’s no clear way to tell whether the current season is just an early start to moderate seasons such as those seen in ’07/’08 and ’10/’11, a late start to a more severe pandemic such as those seen in ’03/’04 and ’09/’10, or something else entirely

Serology offers at least some clue to the severity of the season; the current season is predominantly H3N2 Type A, and historically H3N2 seasons have been nearly twice as severe – and induced substantially more hospitalizations – than seasons in which other serotypes were predominant. Of the 16 H3N2 predominant seasons for which we have estimates of total flu-related hospital admissions, the average estimate (total confirmed and unconfirmed, total season) is roughly 57 admits / 100,000 population, with a range of 26 (’82/’83) to 103 (’97/’98)

By our rough estimates the current season (+/- 36 / 100,000) already has surpassed the low end of the historic H3N2 range. However if the current season were to produce cumulative admits on par with either the average (57) or historic maximum (103), we would expect influenza to produce between +/- 43bp and +/- 2% growth in 1Q13 v. 1Q12 hospital admissions, respectively (Exhibit 2, again)

  1. Thompson et al., “Influenza-Associated Hospitalizations in the United States”, JAMA, September 15, 2004_Vol 292, NO.11, 1333- 1340; and Simonsen et al., “The Impact of Influenza Epidemics on Hospitalizations”, The Journal of Infectious Diseases 2000;181:831-7
Print Friendly