As is the case in many news stories, data analysis can capture and quantify a lot of the "what" but not necessarily the "why." The Las Vegas Sun's series about the state's increasing use of prescription painkillers was no exception.

"The New Addiction" began as a three-part series focusing on the swell of prescription narcotic consumption in Nevada and the rising human toll from the use and abuse of these substances. Analysis of Drug Enforcement Administration reports showed that in 2006 Nevada ranked first in Hydrocodone use per capita - commonly known by brand names like Vicodin and Lortab - and fourth in oxycodone, Methadone and Morphine use. Records obtained from the Clark County coroner showed that the death rate from prescription painkillers had nearly tripled over 10 years and now outpaces fatal overdoses from illicit drugs like Cocaine and methamphetamines.

Our reporting revealed the issue to be every bit as complex as we had expected. The lines among use, abuse and addiction are fuzzy; there is shaky scientific evidence to justify the drugs' exponential expansion beyond treating the terminally ill and/or cancer patients; and there is evidence that the illegal diversion and black market trade of these drugs continues to grow.

But the data work did help us frame our stories, ask better questions and elicit responses from experts concerning our findings.

The bedrock data source for our series came from online reports generated by the DEA's Automation of Reports and Consolidated Orders System (ARCOS) database (www.deadiversion.usdoj.gov/arcos/index. html). The DEA uses it to track the production and distribution of controlled substances to pharmacies and health care practitioners across the country. One inspiration for taking a hard look at the prescription painkillers flowing into Nevada came from reporting and analysis done last year by Frank Bass of The Associated Press.

The ARCOS reports are a gold mine for journalists who want to tackle this issue on their own, but they come with a sizeable caveat: The annual drug summary figures (split by the first three digits of ZIP code) exist as several thousand pages in PDF form. In densely populated metropolitan areas, the generalized ZIP codes may cover a relatively small area, but the ZIPs are much larger in rural areas.

The DEA produces the same information each year in a much shorter collection of annual reports with all substances totaled by state.

The ARCOS reports containing state-level figures showed Nevada as a top consumer of several common prescription painkillers. The DEA already had taken the liberty of calculating a "grams per 100,000 people" rate for each substance and state on the list using Census 2000 population for all years.

Based on that information alone, we knew there had to be an underlying issue here worth examining. We just needed to match the total distribution weight each year with the appropriate population estimates. For certain states with relatively flat growth, using a dated population count wouldn't have much of an effect on where they placed on the list.

But that's not the case for Nevada, one of the fastest growing states in the country. Our population has roughly doubled in the past 15 years and grown by almost 30 percent since the last census. Even accounting for rapid growth, the state still ranked first in the nation for Hydrocodone consumption in 2006, as well as fourth for medical Morphine, Methadone and oxycodone.

Most painkillers are meted out in milligrams (or even micrograms in the case of extremely potent drugs like fentanyl), so we also altered our formula to create a "milligrams per capita" value for each state because we felt it would be an easier number for the reader to digest.

The reports separating drug sales by the first three digits of ZIP code weren't as helpful for the purposes of reporting our story, mainly because of the unbalanced way Nevada's population is distributed. With more than 70 percent of Nevadans concentrated in the Las Vegas metropolitan area and Clark County, the generalized ZIP codes sprawl across the state. It would have been difficult to make statements about the use in particular cities, townships or counties because of the way the data was presented.

But we dug in anyway to create a comprehensive infographie to accompany our series (www. lasvegassun.com/news/2008/jul/06/painful-truthabout-painkillers), focusing on the painkillers most used in Nevada.

To convert the PDFs to a format that lent itself more to analysis, I used Nuance's OmniPage, a commercial optical character recognition (OCR) program. OmniPage allows you to set up a flexible grid of rows and columns on each page for tabular data, as well as to save data in a variety of common electronic formats.

Comma-delimited text files worked best for me; they popped up easily in a single Microsoft Excel sheet for cleaning. Saving the recognized pages as an actual Excel document ended up spreading the data across several hundred sheets in a single workbook. I don't recommend it.

ESRI's ArcView can directly "add data" in the form of sheets from an Excel workbook, so I ne ver had to save the converted ARCOS reports as dBASE files. They imported into the program intact and ready to be joined with a shapefile. I chose a recently updated shapefile that ESRI had provided on CD that already had a recent population estimate for each three-digit ZIP code area.

While the DEA data was great for assessing trends in distribution patterns, it said nothing about the toll of prescription narcotic use. We felt like the data was only part of the puzzle.

Autopsy reports are closed to the public in Nevada thanks to a 25-year-old opinion from Nevada's attorney general. But after the Clark County coroner responded to our polite request, we were able to get access to a simplified internal register that the office uses to track deaths.

We asked them to send a database of all deaths involving prescription painkillers, as well as a database of those involving illicit substances like Cocaine, Methamphetamine and heroin. At no cost to the Sun, the coroner's office was able to provide two Excel spreadsheets that contained nearly 3,700 records, including data such as the case number, deceased's name, date of death and the underlying cause.

I checked the two datasets using Microsoft Access database manager for duplicate records - deaths where both kinds of substances were a factor would appear in both coroner queries - by appending one table to the other and then running a query to look for case numbers that appeared more than once.

I also found that in a handful of cases in the early 1990s, the medical examiner had listed Morphine as a cause of death but later clarified it as heroin. Since it was only a small number, I was able to look at cases where the cause field contained both Morphine and Heroin and determine where they fit in the final analysis.

The databases ended up showing a rapid annual increase in the number of fatal overdoses caused by substances like Methadone and oxycodone. By 2007, deaths caused by these narcotic painkillers had actually overtaken deaths from street drugs. They also eclipsed motor vehicle deaths and were closing in on fatal shootings.

So we felt like we had hard evidence detailing the volume of drugs heading to hospitals, pharmacies and practitioners in our state. And the increase in consumption seemed to be having a dramatic effect on public health. But who was prescribing all of these drugs?

The Nevada Board of Pharmacy carefully guards most of this information. The prescription drug task force tracks prescriptions written and filled for controlled substances. Most states have them these days, but Nevada was one of the first.

In 2007, the database kept information on 260 million doses dispensed in Nevada. Unfortunately, this information included more than just prescription painkillers - there are also many other controlled substances that have medical benefits.

For our story, we were able to get a free extract of the database that contained the number of doses each practitioner had prescribed in the last 1 2 months (all identifying information was removed). The records provided by the state pharmacy board encompassed more than 30,000 prescribers and came in an Excel spreadsheet. It showed that a relatively small number of doctors and physician assistants - 5 percent - were writing prescriptions for nearly 90 percent of controlled substance doses.

All of this analysis was published in a three-part series. The final piece was a follow-up story by Marshall Allen that showed a strong response to our reporting by Nevada legislators. We also invited four members of the medical community and law enforcement to participate in a roundtable discussion on prescription painkillers at the Sun offices. A partial transcript of the discussion was published in the newspaper, with the full roundtable posted on the Web as streaming video.