Restaurant inspections
By Rick Linsk St. Paul Pioneer Press
It took a strong stomach to successfully pull off our "Restaurant Reservations" series (Oct. 23-27, 2000) - not to mention oodles of computer-assisted reporting resources and considerable tenacity.
First, a disclosure: We didn't invent this wheel. Restaurant inspections are a longtime staple of investigative reporting. Television stations in the Twin Cities and other cities, sometimes aided by hidden cameras, have reported filthy or unsafe practices. Newspapers in Toronto, San Jose, Charlotte and St. Louis, among others, have also examined restaurants. Our series built on the techniques of our predecessors, and in some ways went even farther.
We expected to be finished by the end of 1999. But slow-moving government bureaucracies, dirty data, and other factors would confound our timetable.
The editors and reporters central to the project - Projects Editor Jeff Kummer, Food Editor Kathie Jenkins, restaurant beat reporter Gita Sitaramiah and myself, a member of the investigative team - began in mid-summer 1999 with several fundamental questions. We wondered whether area restaurants were generally following or failing the rules; whether the government's system of overseeing restaurants was working; and whether the average consumer had any way to distinguish between the safe and unsafe, the clean and unclean.
We expected to be finished by the end of 1999. But slow-moving government bureaucracies, dirty data, and other factors would confound our timetable.
Getting the records
From the outset, we faced an obstacle: Fragmentation of oversight. Across the nation, restaurants are inspected by a patchwork of state, county and local government health agencies. In Minnesota, ultimate responsibility rests with the state Department of Health, but the department delegates its powers to dozens of counties and cities. In the end, we studied restaurant inspections by MDH and five local agencies in the Twin Cities that were willing and able to give us computerized records.
Some government agencies resisted our request for inspection data and only relented after official requests, meetings, and follow-up phone calls and e-mails. Even then, some took weeks or months to provide the records.
The most notorious agency was the state Department of Health, which oversees food establishments in Dakota County and most of outstate Minnesota. MDH failed to provide complete inspection data for six months, and only after the Pioneer Press noticed that a computer file initially provided by the department was incomplete. The department also took two months to provide computer records about foodborne outbreaks, and only after "sanitizing" the records of information that could identify victims. (More on that below.)
In one county, officials gave us the computer data from inspectors' portable computers, but said their contract with a software vendor prevented them from providing any background on how to read the data.
Analyzing the data
As the old saying goes, be careful what you ask for, because you just might get it. That's how we felt after the various agencies' data arrived. The reason: Regulatory agencies use computers to help keep track of inspection schedules and compliance, not sift trends from the data as we wanted to do. The six databases had several different formats. Some were missing records. Others were littered with duplicate records. One county had only entered about half of all restaurant visits into the computer system - the "bad" inspections, leaving the better visits documented only in hundreds of paper files.
St. Paul's data came from an ancient computer system (finally replaced in the middle of our project) with some confusing twists - among them, taking violations by restaurants long since out of business and attributing them to the building's current occupant. St. Paul also was unable to give us a list of restaurants issued tickets, something other agencies were able to do.
Most of our tasks had to be multiplied by six - for instance, six sets of queries for the worst and best restaurants; six sets of phone calls to officials to request paper records, confirm our findings, ask about "good guys," and other needs; six different sets of preparation materials for interviews with officials, and so forth.
One of the dilemmas we wrestled with was how to analyze the inspection scores. There's a strong argument that scores can be misleading. A restaurant can get a mediocre score, say in the low 80s, for a bunch of minor violations that involve "floor, walls and ceilings" but not food safety. On the other hand, a restaurant can get a score of 90 while committing one or two really gross violations, like improper food temperatures and cross-contaminating cooked food with bacteria from raw food - things that can make people sick.
So from early on, to avoid this controversy, we never contemplated actually publishing the averages, or scores from a given year, etc. Instead, we made plans to count up and rank restaurants based on how many "critical" violations they received.
In total, we found Twin Cities restaurants had violated food-safety rules more than 130,000 times from 1995 to mid-1999, including more than 20,000 violations that government agencies classify as "critical" because they can make diners sick. We also obtained a state database of foodborne illness outbreaks, many of which had never come to light before. As for enforcement, we pieced together agencies' records - sloppy in most cases - and found penalties against violators were rare, even when they were proven to have made people sick.
Tips for others
Advice for this or other health and safety related stories: Carefully document, for instance in a spreadsheet, all your requests for information - when, to whom, how officials responded, and where things stand. It gets harder to keep track as the weeks and months go by, and multiple requests pile up.
To locate victims, consider such public records sources as lawsuits, foodborne illness complaints, death certificates, hospital discharge records, even letters to the editor.
Think early, and if possible decide early, how to document and illustrate your findings regarding specific restaurants. We began by looking at bests and worsts, but had concerns about whether the data would be accurate, timely and complete enough. Late in the game, we shifted gears and wound up publishing inspection results for the most popular restaurants (as measured by the Zagat Survey).
Rick Linsk can be reached by e-mail at rlinsk@pioneerpress.com.