There are 37% fewer providers in Kentucky than in Indiana who prescribed opioid narcotics to Medicare patients in 2013, but they wrote 43% more total opioid prescriptions per prescriber than in Indiana (254 vs. 178). Indiana spent more on opioids overall than Kentucky ($80.1 million vs. $53.6 million) at a cost that averaged 34% higher per prescription. Of the specialties with the largest number of prescribers or the highest rates of individual narcotic prescribing, only Nurse Practitioners and Rheumatologists prescribed a higher rate in Indiana than their Kentucky counterparts. In both states, Nurse Practitioners prescribed opioids at rates substantially below those of Internists or Family Practitioners. The almost across-the-board higher rates in Kentucky were substantial, but were particularly striking for the three major pain-management specialties. Cancer and end-of-life providers in either state did not prescribe opioids at greater than average rates per prescriber– well below those of physicians in general practice and only at a small fraction of the rates of pain-management providers. High-prescribing outliers skewed the averages upwards in several specialties, if not most. The data pose the question: are there structural reasons for the differences in narcotic use between the states, or do the substantial differences result from simple practice preferences or amount of endemic prescription drug abuse?
In a series of articles, I have been exploring the vast database of all outpatient prescriptions written for Medicare Part-D managed care and drug plans which was released by the Centers for Medicare and Medicaid Services. The data enumerates every drug that each named provider prescribed to more than 10 Medicare patients in 2013. With the cooperation of the American Medical Association, drug companies themselves had a good idea who was prescribing what, but this kind of information was previously kept secret from the public. No longer. Fueled by increasing medical costs generally, the exploding cost of drugs, epidemics of prescription drug abuse, and increasing attention to the quality and safety of medical care; detailed provider- and hospital-specific information is being put on the table for all to inspect. (Hopefully medical insurers other than Medicare will be called upon next to tell us what they are doing with our premiums!)
While I continue to believe that examination of such information can identify best medical practices, the reality has been that the worst or frankly illegal prescribing practices are the easiest to find! For example, I find it troubling that the small town of Austin, IN that is staggering under the burden of Opana (oxymorphone) abuse is surrounded by small communities in which Opana is being prescribed in disproportionately large amounts. In my view, until refuted definitively, the medical community itself is fueling this disaster– directly with its prescription pads, and indirectly by lack of adequate professional or regulatory peer review.
When I looked at the prescribing data for Opana in Indiana, I noted that as a group, nurse practitioners prescribed more of that particular drug than any other specialty including family physicians or pain doctors! This can be explained, in my opinion, by high prescription rates by a handful of nurses. To explore further the opioid narcotic prescribing patterns of individual medical specialties, I compared the prescribing patterns in Kentucky and Indiana of all opioids combined by prescriber specialty. As detailed earlier, I extracted from the Medicare database all drugs containing the letter strings ‘cod’ and ‘morph,’ together with fentanyl and meperidine. These identifiers capture all the commonly used opioids prescribed to Medicare patients and no others. (For technical reasons, I excluded the drug Tramadol which was recently re-classified as a Schedule IV opioid-like drug. It was advertised as having low abuse potential but has emerged as a problem drug.) From the data field items provided by Medicare, I calculated the number of unique providers in each specialty, a measure of the number of different opioid preparations used by providers in that specialty, the average number of prescriptions per provider, and the corresponding average costs per prescription and provider. I then ranked and compared the 70 or so different shared Medicare-designated specialties according to the average number of opioid prescriptions by provider within that specialty. Some interesting observations emerged. Summary data across specialties is shown just below. [Links to full data by specialty at end.]
|Rx per Provider||254||178|
|Cost per Provider||$9,328||$8,771|
|Cost per Rx||$37||$49|
Pain doctors write for opioids most frequently.
Not surprisingly, in both states Interventional Pain Management, Pain Management, and Anesthesiology had by far the highest average opioid prescription rates per provider. Physical Medicine & Rehabilitation slid in a little lower at fifth highest. In Kentucky, the average Interventional Pain Management specialist wrote 3,068 opioid prescriptions per year for Medicare patients alone. In Indiana the average was 1,401 for this specialty. A handful prescribed at extraordinary rates. Because there are more providers in these specialties, Internal Medicine, Family Practice, and Nurse Practitioners with relatively high average rates of their own wrote far more prescriptions in aggregate than the three main specialties of pain doctors combined– 884,247 for Kentucky generalists vs. 241,853 for Pain Management providers and Anesthesiologists. In Indiana, the numbers were 1,042,462 vs. 143,693.
Despite the fact that Kentucky had many fewer providers than Indiana that prescribed opioids (5744 vs 9130) it issued nearly the same total number of opioid prescriptions (1.5 vs 1.6 million). Nonetheless, the cost to Indiana was considerably greater than in Kentucky at $80.1 vs. $52.6 million. Individual Kentucky providers provided on average a modestly greater variety of different opioids to their patients.
Kentucky prescribers do it more often.
In general, Kentucky providers of virtually all major specialties who prescribed opioids to more than 10 of their Medicare patients did so on average more frequently than their colleagues across the river in Indiana. A few exceptions make the point that high-prescribing outliers skew upwards the average number of prescriptions per specialty. For example, In Indiana, the average rate of prescribing by Plastic and Reconstructive Surgery specialists was nearly three times the rate for Kentucky providers. On inspection, this difference is largely due to the prescribing habits of a single physician in a small town. Based on the magnitude and scope of the differences, can we conclude that the practice patterns of Kentucky and Indiana prescribers differ? It sure looks that way. Can we conclude that the differences are due to preference rather than medical necessity or appropriateness? Not yet, but that seems likely. If so, our task as professionals goes beyond rooting-out illegal or abusive behaviors by prescribers or patients alike, but must also grapple with the question of whether opioids are being used clinically more than they should. [I think they are.]
Do nurses do it better?
My interest in this area was piqued by the observation that Nurse Practitioners as a group in Indiana prescribed more Opana than even pain doctors. I therefore wanted to look at all opioids combined. As it happens, Nurse Practitioners are the only large specialty in Indiana that wrote prescriptions for opioids at a greater average rate per prescriber than Kentucky. However, as for Opana, their number of prescriptions for opioids was not at the top of the list.
Indiana had 1169 unique Nurse Practitioners prescribing opioids to Medicare patients vs. 684 in Kentucky but the Indiana nurses wrote disproportionally more prescriptions (205,072 vs. 87,181). The rates were 175 prescriptions per provider in Indiana vs. 127 in Kentucky. The cost of nurse-prescribed drugs in Indiana was even more disproportionate– Indiana paid $12.2 million for its Medicare opioids compared to $1.9 million by Kentucky. I conclude that choice of drugs must be correspondingly exotic in Indiana. On inspection, it appears that a large part of these differences was due to the the prescribing habits of a handful of nurses. (I will have more to say later about the substantial variation of prescribing within each specialty in a later article.)
Recall that these data are for Medicare Part-D patients only, although there is little reason to believe that the results are not relevant across the whole payer market. Small numbers of providers in some specialties raise the issue of statistical problems. Some specialties such as Surgical Oncology had only one or two providers prescribing to more than 10 Medicare patients. Statistically, apparent differences between such small groups can occur more easily by chance alone.
Note that Medicare does not always get the specialty right. A few of those listed would not be licensed to prescribe narcotics at all and likely reflect administrative errors. Other errors in assembling the database must certainly be present. It is also certain that some providers in specialties such as Internal Medicine, Family Practice, Nurse Practitioner and others may well emphasize pain-management in their practices. The practice settings of these providers is unknown and may include nursing homes or hospice units. Recall that Medicare patients fall into two major groups– the elderly older than 65, and the qualifying disabled of all ages. I suspect that there are more disabled Medicare patients in Kentucky– at least Eastern Kentucky. I believe that disability and prescription drug abuse are correlated in Kentucky and elsewhere.
It may be that licensing or other regulations in the two states, such as the prescribing privileges of nurses, may be operative. Does anyone have information about this or any other considerations that might help us understand the data better?
I was very surprised to see so many opioids prescribed by Rheumatologists in both states. Their opioid prescription rates were near the top. Either best practices have changed since I taught medicine and was in practice, or previous rheumatologists were using opioids secretly! In Kentucky it is difficult to understand why Allergy/Immunology ranked #13 in prescription rate, but this also appears to be due, to a single provider outlier. I was surprised both by the relatively low rate of opioid prescriptions by Hospice and Palliative Care providers in general, and the large difference between the two states in this specialty. Kentucky cardiologists seem to like using opioids more than their colleagues across the river. I have not yet broken down all the specialties to look at individual prescriber profiles.
Despite all the talk about how important, indeed essential, opioids are for the care of cancer patients, the prescription rates for the several oncology specialties in both states were above the median but much below those of the general practice specialties and not much different from other general medical and surgical sub-specialties that also care for some cancer patients.
It is easy to focus on pain-management providers as we grapple with prescription drug abuse and diversion. However, the data clearly show that the large majority of prescription narcotics are prescribed by generalists such as Internists, Family Physicians and Nurse Practitioners. If we are serious about establishing best practices for narcotic use, we must shine a bright light on these groups as well. Some providers will feel under attack, but we cannot all be right all the time. Surely as a group we can practice better and more responsibly.
I must also unload a nagging thought here that I cannot dismiss. I have always believed that all doctors are pain doctors. That is what has always brought patients into our offices. The use of opioid narcotics by pain doctors, interventional or not, is stunningly higher than that of other medical providers including those who care for cancer and end-of-life patients. Is it possible, that because of all the heavy lobbying to consider pain as the 5th vital sign, and that by allowing physicians to self-differentiate themselves as “pain doctors” that we have as a profession and community given these new professional groups immunity to overprescribe without consequence? Is it appropriate to allow only other pain doctors to pass judgment on each other. I do not believe so. Competent physicians and providers of all specialties share a common background and have a voice that deserves to be heard. Mine is that opioid narcotics are vastly over-promoted and overused in American medicine. Please convince me that I am wrong.
[Notice: An earlier version of this article was posted August 27 but withdrawn the next day when I realized that the drug tramadol had been inadvertently included in the drugs studied in Kentucky but not Indiana. The data has been recalculated without tramadol in the article above and in all the tables and spreadsheets. Not surprisingly, the overall number of prescribers, prescriptions, costs, and prescriptions per prescriber fell modestly for Kentucky. The cost per prescription actually rose from $ 34 to $37. However, no conclusion or opinion is appreciably different in this revision.]
Peter Hassselbacher, MD
Emeritus Professor of Medicine, UofL
August 30, 2015
I have attached three documents that present the comparisons above in greater detail as both PDF documents and as Excel spreadsheets. The first two are relatively straight-forward if not over-abundantly filled with numbers. For each of the two states, I list all the specialties identified by Medicare populated by individual providers prescribing to more than 10 Medicare patients. The master database has multiple lines for each unique provider, one for each different drug prescribed to more than 10 patients. The number of original “Line Items” for any specialty is generally more than the number of unique prescribers. I therefore aggregated all opioid prescriptions for each unique prescriber into a single data row. Dividing the total line items by the number of unique providers gives an index of how many different drugs are used in each specialty.
From the number of prescriptions and refills and the total cost provided by Medicare, I also calculated the number of prescriptions per provider, the cost per provider, and the cost per prescription. The data are ranked by total numbers of prescriptions and refills per specialty group. The descriptive statistics at the bottom of the pages show the aggregate grand totals for Prescribers, Prescriptions, and Cost and then compare the average and median values of each specialty.
The third attachment ranks the listed specialties prescribing opioids for each state ranked primarily by the number of prescriptions per provider in each group. Because some specialties were represented in one state but not the other, and because the ranking order was not exactly the same, I paired the specialty for Indiana next to the corresponding one for Kentucky. The column on the far right calculates the differences between the ranks and provides an index of when there is a substantial difference between the two states. Note that the major populated specialties and top prescribers track each other fairly well.
If I have made an error of fact or interpretation, or you have additional insights, let me know in the public comments section below, or privately.
• Prescription of opioids to Medicare patients of KY by specialty, 2013. (PDF) (Excel).
• Prescription of opioids to Medicare patients of IN by specialty, 2013. (PDF) (Excel).
• Comparison of prescription of opioids KY to IN by specialty, 2013, (PDF-3pg) (PDF-1pg Tabloid) (Excel).