Health Data Miner

Two research firms in the healthcare IT space have produced press releases in recent weeks predicting the next "great wave" in EHR purchases among U.S. hospitals.

[See also: Consider the power of incentives]

"One in three large hospitals are in a buying spree," reports Black Book Market Research, "selecting new EHRs, changing out original systems after implementation delays, cost run-ups and physician revolts, meaningful use snafus, silo complaints, interoperability disconnects, and accountable care reforms."

Meanwhile, KLAS issued a release, "Acute Care EMR 2014: The Next Buying Wave," that begins: "Almost half of large hospitals interviewed indicated that they will be making a new EMR purchase by 2016…"

[See also: EHR market forecast at $5.4B by 2015]

"Where the last round of EMR purchases was fueled by meaningful use requirements and enticing reimbursements, this next round is being fueled by concerns about outdated technology and health system consolidation," said report author Colin Buckley. "This shift in focus will play a major factor in which EMRs are being considered."

Wow! Makes you wish Epic were publicly traded, huh?

But to be honest, I'm not in a hurry to rearrange my (modest) portfolio just yet. While the data I'm looking at doesn't rule out a surge in EHR acquisition, it provides some reason for skepticism.

The main reason is historical activity. The chart below shows the install and replacement rate of clinical data repositories (a proxy for EMRs) according to the HIMSS Analytics Database and Dorenfest Institute.

Over the past five years, the average replacement and installation rate of EHRs is 10.7 percent annually for hospitals with 200 or more beds. If you included all hospitals in the HA database, the replacement and installation rate drops to just 3.8 percent.

I asked Lorren Pettit, vice president of market research at HIMSS Analytics, if the historical data provided any clues about future activity in the space. Using regression analysis, Pettit says the data predict "a slight decline in the percentage of hospitals each year replacing their" EHRs – about .6 percent annually.

So what explains the apparent disparity between the HIMSS Analytics Database, and the KLAS and Black Book results?

Partly, you need to read the fine print. For instance, in the Black Book Market Research release, an explanatory paragraph notes that 32 percent of large hospitals are reevaluating their vendors, but only 19 percent indicate that the reevaluation "will likely lead to a replacement system."

So what the Black Book report really says is that only one in five large hospitals participating in the survey are "likely" to replace their EHRs. Which in turn begs the questions, "How likely?" and, "When?" Since EHR replacement projects at large hospitals can take several years to complete, even the 19 percent that are likely to replace their EHRs won't do so in the same year.

That may also explain some of the KLAS findings. Assuming KLAS conducted the research in mid- to late-2013, finding that "nearly half" of large hospitals will replace EHRs by the end of 2016 would mean that about 14 percent of large hospitals plan replacement each year.

That's a 40 percent higher annual replacement rate than the HIMSS Analytics data suggest, or about 64 more hospitals a year (based on 1,600 hospitals with more than 200 beds).

Another explanation for the difference in perspectives might be the nature of the data itself. KLAS' methodology includes sampling the industry, and it provides opportunities for self-reporting. Thus, potential for sample bias exists.

Think about the 1948 election, and the Chicago Tribune headline that read "Dewey Defeats Truman." Truman actually won, but the newspaper made its erroneous prediction based on early election returns and polls that appeared to show an inevitable Dewey victory. But the early returns weren't representative samples, and the polls' samples were biased because they relied on telephone surveys – people without phones weren't represented in the sample.

It could be that hospitals that have "concerns about outdated technology and health system consolidation," as the KLAS report noted, were more likely to participate in the study. This would weight the sample in favor of hospitals with purchasing plans, since unhappy CIOs are more likely to invest in replacement systems than CIOs who are content with their EHR.

What conclusions do I draw? All three research firms seem to predict that between 8 and 14 percent of large hospitals will be buying new EHRs in each of the coming years. The difference is significant, but not extraordinary.

More importantly, the range is consistent with HIMSS Analytics historic data for the past five years (a range of 8.2 percent in 2013 to 14 percent in 2008). So I don't see a surge, a spree or even a wave, in this case – just business as usual.