Two-year-old Mikey Fernandez bumped his head, and his parents took him to Rush Presbyterian Hospital in Chicago as a precaution. He was given a sedative before a CAT scan. The nurse gave him an overdose of the sedative, and the child died.

A California woman was given Pitocin to induce labor. The hospital miscalculated the dosage, sending 35 times the proper dose into her bloodstream, killing her unborn daughter.

These were two of the 1,720 hospital patients accidentally killed (9,584 others were injured) over a five-year period by the "actions or inaction of registered nurses besieged by cuts in staff and other belt-tightening in U.S hospitals," the Chicago Tribune reported.

The findings were made possible by more than a dozen state and federal databases and old-fashioned shoe-leather reporting.

Useful Databases

Mike Berens, a Tribune reporter who worked on the story, said two databases "proved most useful in tracking nursing errors":

MAUDE—Compiled by the U.S. Food and Drug Administration; contains more than 3 million records detailing possible medical device failures. It was used to find instances of human error. "Like most computer databases, MAUDE data was a beginning for my reporting, not an end," Berens said.

List of disciplined nurses—Kept by the National Council for State Boards of Nursing; includes nurse's identity and type of violation. The council does not allow public access, but mails files to each state. "That means the data become part of the public record," Berens says.

The information from these two databases and from other public records was cross-matched. Berens found: "Files documented that drug-addicted nurses roamed from hospital to hospital without punishment. Nurses with serious felony convictions—from child molesting to aggravated drug-trafficking—continued to work with impunity."

From his first interviews with nurses, Berens was told that unprecedented staffing cutbacks were endangering hospital patients. The American Hospital Association replied that a record number of registered nurses were at work. Berens then entered on a spreadsheet the association's data and cross-matched it with information about staffing levels he found in nurse union contracts and Medicaid cost reports that hospitals are required to file with the U.S. Department of Health and Human Services.

"My analysis clearly showed that fewer nurses were assigned to patient care. My conclusion: The AHA data was misleading and flawed," Berens said. The AHA then admitted its list included many nurses who were desk-bound or assigned to non-hospital nursing homes and home health agencies.

Berens' advice, which can be found in the Nieman Reports Winter 2000:

The computer provides reporters access to information that can't be seen in any other way. Increasingly, public information is stored exclusively in computers. Those who don't know how to enter these electronic vaults remain modem-day illiterates. Would anyone hire a reporter who refused to use the telephone? Yet, in the early days of this groundbreaking technology, there were undoubtedly those who argued that face-to-face information gathering was the only proper way to do the job.

Fundamental reporting skills are still essential. A computer will not make a bad reporter good. But the computer empowers journalists to go beyond the press release. Few interview techniques are more potent or fruitful than confronting officials with computer analysis of their data.

Other CAR Stories

Examples of computer-assisted reporting appear in newspapers and on television every day. You can track them by visiting The Scoop [http://www.thescoop.org/], a Weblog maintained by Derek Willis, research database editor at The Washington Post. Willis posts summaries of such stories, with links to the full articles.

Here are some CAR examples cited by Willis in late 2004:

• Robert Anglen of The Arizona Republic analyzed police reports of Taser-related incidents in 2003 and "found that Phoenix police were far more likely to use the stun guns to make someone obey orders at a traffic stop than to bring down an armed robber." More than 5,000 police agencies, including 108 in Arizona, have armed their officers with Tasers. "The Republic's analysis of use-of-force records shows that in nearly nine out of 10 cases, individuals did not threaten police with weapons."
  
  http://www.azcentral.com/specials/special43/articles/1107tasermain07.html



• Barbara Clements and David Wickert of The News Tribune in Tacoma, Wash., used state and federal records to investigate safety in the construction industry: "Three construction workers die every day in the United States, with about a third of those falling from heights, making falls the chief cause of death among construction workers in any given year. In Washington state, it's the same story. A construction worker is more likely to fall to his death than any other cause, a review of state data since 1998 shows. State and national officials have launched safety programs, strengthened fines and passed new laws, but none of these actions has dropped the death toll for residential and commercial construction workers from about 1,100 a year for the last decade." Overworked inspectors, weak enforcement and the number of immigrants who work in the industry were cited as factors.
  
  http://www.thenewstribune.com/business/story/4273644p-4068260c.html



• Sally Kestin and Megan O'Matz of the South Florida Sun-Sentinel found that the Federal Emergency Management Agency "approved $28,000,000 in storm claims" in Miami-Dade County, 100 miles south of where Hurricane Frances landed. The claims were "for new furniture and clothes and thousands of new televisions, microwaves, refrigerators and other appliances." The series of reports revealed that FEMA "paid for new cars, dental bills and a funeral, even though the Medical Examiner recorded no deaths from Frances."
  
  http://www.sun-sentinel.com/news/sfl-femacoverage,0,6697347.storygallery



• During the November 2006 elections, Duane Pohlman, an investigative reporter at WEWS-TV in Cleveland, compared Cuyahoga County voting records with death records provided by the Social Security Administration. Dozens of dead people have had votes cast in their names. U.S. Rep. Dennis Kucinich called this an "an assault on the integrity on the electoral process."
  
  http://www.newsnet5.com/investigations/10209073/detail.html



• Ken Ward Jr. of The Charleston Gazette in West Virginia analyzed government reports and data on deaths in coal mines nationwide over the past 10 years. He found that nine of every 10 coal mining deaths could have been avoided if existing safety rules had been followed.
  
  http://www.sundaygazettemail.com/section/News/2006110421

Investigative Reporters and Editors Inc. also compiles examples of CAR. IRE has a Weblog similar to The Scoop. It's called Extra!Extra! [http://www.ire.org/extraextra/]. In addition, the association's members can access the IRE Resource Center, which contains abstracts of more than 20,000 investigative stories—many of which used CAR. The Resource Center also has tipsheets on the gamut of journalism skills, including CAR.

The Computer-Assisted Paragraph

David Donald, training director for the Investigative Reporters and Editors Inc. and the affiliated National Institute for Computer-Assisted Reporting, said it's getting easier to do CAR: "More and more agencies are making the reams of data they collect available through the Web. And when reporters find those data sets on the Internet, download them and put them into a spreadsheet or database manager for analysis, the power of the Web becomes obvious."

As a reporter for The Charleston Daily Mail, Brian Bowling has broken a lot of big stories about mining violations and other topics. But it's not just projects that benefit from CAR, he said. "Reporters with access to the tools can use just a ‘little' CAR to improve an otherwise routine story."

A "computer-assisted paragraph" can add authority and depth to such a story, putting the news in context. In writing about a murder, for instance, a savvy reporter might do a quick analysis of crime and census data to describe homicide trends and the neighborhood where the killing happened.