Changes in Opioid Prescribing in the United States, 2006–2015

Vital Signs: Changes in Opioid Prescribing in the United States, 2006–2015

Gery P. Guy Jr., PhD1; Kun Zhang, PhD1; Michele K. Bohm, MPH1; Jan Losby, PhD1; Brian Lewis2; Randall Young, MA2; Louise B. Murphy, PhD3; Deborah Dowell, MD1 (View author affiliations)

View suggested citation and related materials

Key Points

• The amount of opioids prescribed in the United States peaked in 2010 and then decreased each year through 2015. Despite reductions, the amount of opioids prescribed remains approximately three times as high as in 1999.

• Opioid prescribing varied substantially across the country, with average per capita amounts prescribed in the top-prescribing counties approximately six times the amounts prescribed in the lowest prescribing counties in 2015.

• Higher amounts of opioids were prescribed in counties with a larger percentage of non-Hispanic whites; a higher prevalence of diabetes and arthritis; micropolitan counties; and counties with higher rates of unemployment and Medicaid enrollment.

• The substantial variation in opioid prescribing observed at the county-level suggests inconsistent practice patterns and a lack of consensus about appropriate opioid use and demonstrates the need for better application of guidance and standards around opioid prescribing practices.

• Health care providers can follow the CDC’s Guideline for Prescribing Opioids for Chronic Pain, which provides evidence-based recommendations about opioid prescribing for primary care clinicians treating adult patients with chronic pain, outside of active cancer treatment, palliative care, and end-of-life care.

• Additional information is available at https://www.cdc.gov/vitalsigns/.

Abstract

Background: Prescription opioid–related overdose deaths increased sharply during 1999–2010 in the United States in parallel with increased opioid prescribing. CDC assessed changes in national-level and county-level opioid prescribing during 2006–2015.

Methods: CDC analyzed retail prescription data from QuintilesIMS to assess opioid prescribing in the United States from 2006 to 2015, including rates, amounts, dosages, and durations prescribed. CDC examined county-level prescribing patterns in 2010 and 2015.

Results: The amount of opioids prescribed in the United States peaked at 782 morphine milligram equivalents (MME) per capita in 2010 and then decreased to 640 MME per capita in 2015. Despite significant decreases, the amount of opioids prescribed in 2015 remained approximately three times as high as in 1999 and varied substantially across the country. County-level factors associated with higher amounts of prescribed opioids include a larger percentage of non-Hispanic whites; a higher prevalence of diabetes and arthritis; micropolitan status (i.e., town/city; nonmetro); and higher unemployment and Medicaid enrollment.

Conclusions and Implications for Public Health Practice: Despite reductions in opioid prescribing in some parts of the country, the amount of opioids prescribed remains high relative to 1999 levels and varies substantially at the county-level. Given associations between opioid prescribing, opioid use disorder, and overdose rates, health care providers should carefully weigh the benefits and risks when prescribing opioids outside of end-of-life care, follow evidence-based guidelines, such as CDC’s Guideline for Prescribing Opioids for Chronic Pain, and consider nonopioid therapy for chronic pain treatment. State and local jurisdictions can use these findings combined with Prescription Drug Monitoring Program data to identify areas with prescribing patterns that place patients at risk for opioid use disorder and overdose and to target interventions with prescribers based on opioid prescribing guidelines.

 Top

Introduction

In 2015, drug overdoses accounted for 52,404 deaths in the United States, 63.1% of which involved an opioid (1). Among opioid-related deaths, approximately 15,000 (approximately half) involved a prescription opioid (2). In addition, an estimated 2.0 million persons in the United States had opioid use disorder (addiction) associated with prescription opioids in 2015 (3). The economic burden of prescription opioid overdose, abuse, and dependence is estimated to be $78.5 billion each year in the United States (4). Prescription opioid-related overdose deaths and admissions for treatment of opioid use disorder have increased in parallel with increases in opioids prescribed in the United States, which quadrupled from 1999 to 2010 (5). This increase was primarily because of an increase in the use of opioids to treat chronic noncancer pain (6,7). Previously, opioids had primarily been reserved for severe acute pain, postsurgical pain, and end-of-life care. This change in prescribing practice increased the amount of opioids prescribed for three reasons. First, opioid use for chronic noncancer pain increased the number of opioid prescriptions. Second, the use of opioids to treat ongoing chronic conditions increased the average lengths of time for which opioids were prescribed (6,7). Third, average dosages of opioid prescriptions tend to be higher for patients who are prescribed opioids for long periods of time, effectively increasing the average amount of opioids supplied per prescription (6,7). Together, these changes placed more persons at risk for opioid use disorder and overdose (811).

Chronic pain is one of the most common reasons for seeking medical attention in the United States, and prescription opioids are frequently prescribed to manage pain (12). However, opioids should only be used when benefits are expected to outweigh risks. Ensuring that patients have access to safe, effective treatment is critical and involves improving the way opioids are prescribed. To improve understanding of opioid prescribing trends in the United States before the release of CDC’s 2016 Guideline for Prescribing Opioids for Chronic Pain (Guideline), CDC analyzed changes in national and county-level opioid prescribing and characteristics associated with higher prescribing rates at the county-level (13).

 Top

Methods

Data on opioid prescribing come from the QuintilesIMS Transactional Data Warehouse, which provides estimates of the number of opioid prescriptions dispensed in the United States based on a sample of approximately 59,000 pharmacies, representing 88% of prescriptions in the United States.

Changes in opioid prescribing at the national level were analyzed from 2006 to 2015. Prescribing rates included overall opioid prescribing rates, high-dose prescribing rates, and prescribing rates by days’ supply (≥30 days and <30 days). Annual opioid prescribing rates were calculated by dividing the number of opioid prescriptions by the U.S. Census population estimates each year. High-dose prescribing rates include prescriptions with daily dosage ≥90 morphine milligram equivalents (MME) (13). All rates are per 100 persons. Additional measures included MME per capita, average daily MME per prescription, and average days’ supply per prescription. Cold and cough products containing opioids and buprenorphine products indicated for conditions other than pain were excluded.

To determine where prescribing changes occurred, opioid prescribing at the county level was examined in 2010 (when prescribing levelled off nationally) and 2015. Quartiles were created using MME per capita to characterize the distribution of opioids prescribed. The percentage of counties experiencing changes in opioid prescribing measures from 2010 to 2015 was calculated. A change of ≥10% was considered to be an increase or decrease, whereas changes <10% were considered stable. County-level characteristics were examined in 2015 by MME per capita quartiles. County characteristics were obtained from the U.S. Census Bureau (age, urban/rural status); American Community Survey (race/ethnicity, percent uninsured, percent unemployed, income); U.S. Diabetes Surveillance System (diabetes prevalence); Dartmouth Atlas of Health Care (provider supply); Centers for Medicare and Medicaid Services (Medicaid and Medicare coverage); Behavioral Risk Factor Surveillance System (arthritis prevalence); and the Area Health Resource File (percent disabled, suicide rate). To identify county-level factors associated with MME per capita in 2015, a stepwise multivariable linear regression model incorporating age, race/ethnicity, insurance status, education, unemployment rates, poverty rates, median income, urban/rural status (metropolitan, micropolitan [i.e., town/city; nonmetro], and noncore [i.e., rural; nonmetro]), suicide rates, dentist and primary care physician density, and diabetes, arthritis, and disability prevalence was estimated.

 Top

Results

In the United States, annual opioid prescribing rates increased from 72.4 to 81.2 prescriptions per 100 persons from 2006 to 2010, were constant from 2010 to 2012, and then decreased by 13.1% to 70.6 per 100 persons from 2012 to 2015 (Figure 1). Annual high-dose opioid prescribing rates remained stable from 2006 to 2010 and then declined by 41.4% from 11.4 per 100 persons in 2010 to 6.7 in 2015. Annual prescribing rates for prescriptions of ≥30 days’ supply increased 58.9% from 17.6 per 100 persons in 2006 to 28.0 per 100 persons in 2012 and leveled off from 2012 to 2015. Annual prescribing rates for prescriptions of <30 days’ supply were stable from 2006 to 2012 and decreased 20.2% from 53.2 per 100 persons in 2012 to 42.4 in 2015. Average daily MME per prescription remained stable from 2006 to 2010 and then decreased 16.9% from 58.0 in 2010 to 48.1 in 2015. Average days’ supply prescribed increased 33.0% from 13.3 in 2006 to 17.7 in 2015.

From 2010 to 2015, the amount of opioids prescribed in the United States decreased from 782 to 640 MME per capita (data not shown). In 2010 and 2015, the amount of opioids prescribed across counties varied substantially (Figure 2). From 2010 to 2015, among counties with sufficient data MME per capita decreased in 49.6% of counties, remained stable in 27.8% of counties, and increased in 22.6% of counties (Table 1). Overall prescribing rates decreased in nearly half (46.5%) of counties, whereas high-dose opioid prescribing rates and average daily MME per prescription decreased in the majority of counties, with 86.5% and 72.1% of counties, respectively, experiencing decreases. From 2010 to 2015, average number of days’ supply increased in 73.5% of counties.

Despite reductions in prescribing, the amount of opioids prescribed in 2015 remained high relative to 1999 levels and varied substantially across the country, from an average of 203 MME per capita in the lowest quartile to 1,319 MME per capita in the highest quartile. Opioid prescribing amounts varied across several county-level characteristics (Table 2). After adjustment in the multivariable model, the following characteristics were associated with higher amounts of opioids prescribed: a larger percentage of non-Hispanic whites; higher rates of uninsured and Medicaid enrollment, lower educational attainment; higher rates of unemployment; micropolitan status; more dentists and physicians per capita; a higher prevalence of diagnosed diabetes, arthritis, and disability; and higher suicide rates. Together, these factors explain approximately 32% of the variation in the amount of opioids prescribed at the county-level.

 Top

Discussion

The amount of opioids prescribed in the United States began to decrease in 2011. However, in 2015, at 640 MME per capita, it remains approximately three times as high as in 1999, when 180 MME per capita were sold in the United States (5), and nearly four times as high as the amount distributed in Europe in 2015 (14).

Two prescribing changes appear to be associated with the decrease in MME prescribed per capita in the United States from 2010 to 2015. First, average daily MME per prescription decreased after 2010, both nationwide and in most counties. The largest decreases occurred from 2010 to 2012, following publication of two national guidelines defining high-dose opioid prescribing as >200 MME/day (15,16). It also coincided with studies demonstrating progressively increasing overdose risk at prescribed opioid dosages exceeding 20, 50, and 100 MME per day (911) and publications highlighting associations of prescribed opioids with overdose deaths (5,17). Second, the rate of opioid prescribing decreased nationwide and in many counties. Nationally, opioid prescribing rates leveled off from 2010 to 2012, and then decreased by 13.1% from 2012 to 2015. These decreases might reflect growing awareness among clinicians and patients of the risks associated with opioids. Throughout this period, however, the average duration of opioid prescriptions increased, in part because of the continued increase in longer opioid prescriptions (≥30 days) through 2012, followed by a stabilization of the rate, and a substantial decrease in shorter prescriptions (<30 days) after 2012. This pattern, along with the trends in overall numbers of opioid prescriptions, might reflect fewer patients initiated on opioid therapy after 2012, whereas patients already receiving opioids were more likely to continue receiving them. Patients are at risk for continuing opioids long-term once they have received them for >5 days (18), and are unlikely to discontinue opioids after they have received them for 90 days (19), highlighting both the importance of minimizing unnecessary initial opioid exposure and potential challenges in reducing opioid use among patients already receiving them.

From 2010 to 2015, half of counties in the United States experienced reductions in the amount of opioids prescribed, with substantial decreases in certain states. In 2011 and 2012, Ohio and Kentucky, respectively, mandated that clinicians review Prescription Drug Monitoring Program (PDMP) data and implemented pain clinic regulation (20). MME per capita decreased in 85% of Ohio counties and 62% of Kentucky counties from 2010 to 2015. In Florida, where multiple interventions targeted excessive opioid prescribing from 2010 through 2012, (e.g., pain clinic regulation and mandated PDMP reporting of dispensed prescriptions) (21), the amount of opioids prescribed per capita decreased in 80% of counties from 2010 to 2015. During this time, Florida also experienced reductions in prescription opioid-related overdose deaths (21).

Despite reductions, the amount of opioids prescribed in 2015 remained high relative to 1999 levels and varied substantially across the country, with average per capita amounts prescribed in the top quartile of counties approximately six times the amounts prescribed in the lowest quartile. Larger amounts were prescribed in micropolitan counties and in counties with a higher prevalence of diagnosed diabetes and arthritis. The latter finding might represent treatment for pain associated with these or co-occurring painful conditions. However, there are effective nonopioid treatments for pain whose benefits outweigh the harms (13). Reasons for higher opioid use in micropolitan counties might include less access to quality health care and other treatments for pain, such as physical therapy. In addition, persons in rural areas might travel to micropolitan areas, which often serve as an anchor community for a much larger rural region, to receive medical care and pick up medications.

Despite reductions in opioid prescribing in recent years, opioid-involved overdose death rates continue to increase. However, these increases have been driven largely by use of illicit fentanyl and heroin (1). There is no evidence that policies designed to reduce inappropriate opioid prescribing are leading to these increases. Combined implementation of mandated provider review of PDMP data and pain clinic laws reduced the amount of opioids prescribed, prescription opioid-involved overdose deaths, and all opioid-involved deaths (20). The policies were also associated with reductions in heroin overdose deaths that were not statistically significant (20). By reducing the number of persons exposed to opioids and the subsequent risk of opioid use disorder these policies might reduce the number of persons initiating illicit opioid use in the longer term (20).

The findings in this report are subject to at least four limitations. First, QuintilesIMS estimates of dispensed prescriptions have not been validated, and they do not include prescriptions dispensed directly by prescribers (although this likely represents a small minority of prescribed opioids), potentially biasing opioid prescribing downwards. Second, county-level analyses are aggregated by the county where an opioid is dispensed, and cannot account for prescriptions obtained by persons outside of the county. Third, the analysis does not include clinical outcomes. However, previous analyses have found associations between population-level amounts of opioids prescribed and opioid overdose death rates (5), and between prescribed dosages and individual overdose risk (911). Finally, because data on the indications for which opioids were prescribed were not available, the appropriateness of opioid prescriptions, or whether opioids were prescribed for acute, chronic, or end-of-life pain, could not be determined.

Although some variation in opioid prescribing is associated with characteristics such as the prevalence of possibly painful conditions (e.g., arthritis), differences in these characteristics explain only a fraction of the wide variation in opioid prescribing across the United States. This variation suggests inconsistent practice patterns and a lack of consensus about appropriate opioid use and demonstrates the need for better application of guidance and standards around opioid prescribing practices (13). CDC’s Guideline provides evidence-based recommendations about opioid prescribing for primary care clinicians treating adult patients with chronic pain outside of active cancer treatment, palliative care, and end-of-life care (13). The Guideline can help providers and patients weigh the benefits and risks for opioids according to best available evidence and individual patients’ needs and safely taper opioids if risks outweigh benefits. The Guideline recommends the use of nonopioid therapies, such as acetaminophen, nonsteroidal anti-inflammatory medications, exercise therapy, and cognitive behavioral therapy for chronic pain (13).

Given associations between opioid prescribing, opioid use disorder, and opioid overdose rates (5), states and local jurisdictions can use these findings to target high-prescribing areas for interventions such as academic detailing for clinicians or individual educational visits to clinicians (22), and increased access to medication-assisted treatment for patients with opioid use disorder. Innovative approaches such as virtual physical therapy sessions with pain coping skills training (23,24) can be used to improve access to effective treatment for chronic pain. In addition, states can consider policies that can reduce opioid overdose, including mandated PDMP use and pain clinic laws (20). Changes in opioid prescribing can save lives. The findings of this report demonstrate that substantial changes are possible and that more are needed.

 Top

Acknowledgments

Puja Seth, PhD; Rose Rudd, MSPH; Lyna Schieber, DPhil; Felicita David, MS, National Center for Injury Prevention and Control, CDC.

Prescription Opioid Use, Misuse, and Use Disorders in U.S. Adults

Prescription Opioid Use, Misuse, and Use Disorders in U.S. Adults: 2015 National Survey on Drug Use and Health


Reviews

Treatment of Latent Tuberculosis Infection  FREE  VIDEO


ADVERTISEMENT

 


Beyond the Guidelines

Should This Patient Receive an Antidepressant?  CME/MOC  VIDEO


Editorials

Prescription Opioid Use Among U.S. Adults: Our Brave New World


On Being a Doctor

Crutches  AUDIO


Observations

Immune Checkpoint Inhibitor Therapy in a Liver Transplant Recipient With Melanoma


In the Clinic

Osteoporosis  CME/MOC


Web Exclusives

Annals Graphic Medicine – American Doctors  FREE

Annals for Educators – 1 August 2017  FREE


Summaries for Patients

Prescription Opioid Use, Misuse, and Use Disorders in U.S. Adults  FREE


Click here to set up a free account or manage your Annals Email Alerts

Visit Annals.org

 


Cravings for opioids, narcotics, higher in women than men – higher in alcohol cravings

The majority of men and women tested positive for oxycodone (68% and 65%, respectively) and morphine (89% each).

More women than men tested positive for amphetamines (4% vs. 1%, p<0.01), methamphetamine (11% vs. 4%, p<0.01) and phencyclidine (8% vs. 4%, p=0.02). More men than women tested positive for methadone (11% vs. 6%, p=0.05) and marijuana (22% vs. 15%, p=0.03).

Craving for opioids was significantly higher among women (p<0.01).

Men evidenced higher alcohol (p<0.01) and legal (p=0.04) ASI composite scores, whereas women had higher drug (p<0.01), employment (p<0.01), family (p<0.01), medical (p<0.01), and psychiatric (p<0.01) ASI composite scores. Women endorsed significantly more current and past medical problems.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3164783/


Rates of lifetime and past-year non-medical use of prescription opiates were 13.6% and 5.1%, respectively. Significantly more men than women endorsed lifetime (15.9% vs. 11.2%) and past-year use (5.9% vs. 4.2%; ps<0.0001). Among past-year users, 13.2% met criteria for current prescription opiate abuse or dependence, and this did not differ significantly by gender.

Polysubstance use and treatment underutilization were common among both men and women, however significantly fewer women than men had received alcohol or drug abuse treatment (p=0.001).

Men were more likely than women to obtain prescription opioids for free from family or friends, and were more likely to purchase them from a dealer (ps<.01). Gender-specific predictors of use as compared to abuse/dependence were also observed.

http://www.ncbi.nlm.nih.gov/pubmed/20598809/


Despite the fact that important gender differences in drug and alcohol use have been previously reported, little research to date has focused on gender differences with regard to nonmedical prescription opioid use. This study preliminarily examined the presenting characteristics and correlates (e.g., age of onset, route of administration, motives for using, and method of introduction) of men and women with prescription opioid dependence. Participants were 24 (12 men and 12 women) non-treatment seeking individuals at least 18 years of age with current (i.e., past 12 months) prescription opioid dependence who participated in an in-depth interview. The average age of onset of prescription opioid use was 22.2 years (SD=8.5). In comparison to men, women were approximately six years older when they initiated prescription opioid use, but were only three years older when they began to use prescription opioids regularly (i.e., weekly), suggesting an accelerated course of disease progression among women. Over half of the sample (61.5%) endorsed chewing and almost half (45.8%) endorsed crushing and snorting prescription opioids. Men were significantly more likely than women to crush and snort prescription opioids (75.0% vs. 16.7%; p=0.01).

Women were significantly more likely than men to be motivated to use prescription opioids in order to cope with interpersonal stress, and to use them first thing in the morning (ps=0.04). Concomitant alcohol and other drug use were common among both men and women. The findings highlight clinically relevant gender differences and may help enhance the design of gender-sensitive screening and treatment interventions for prescription opioids.

http://www.ncbi.nlm.nih.gov/pubmed/21514061


29,906 assessments from 220 treatment centers were included, of which 12.8% (N=3821) reported past month prescription opioid abuse. Women were more likely than men to report use of any prescription opioid (29.8% females vs. 21.1% males, p<0.001) and abuse of any prescription opioid (15.4% females vs. 11.1% males, p<0.001) in the past month. Route of administration and source of prescription opioids displayed gender-specific tendencies.

Women-specific correlates of recent prescription opioid abuse were problem drinking, age <54, inhalant use, residence outside of West US Census region, and history of drug overdose. Men-specific correlates were age <34, currently living with their children, residence in the South and Midwest, hallucinogen use, and recent depression.

Women prescription opioid abusers were less likely to report a pain problem although they were more likely to report medical problems than women who abused other drugs.

http://www.ncbi.nlm.nih.gov/pubmed/19409735

Victim blaming

Victim blaming (or blaming the victim) is holding the victim (s) of a crime, an accident, or any type of abusive maltreatment to be entirely or partially responsible for the transgressions committed against them. Victim-blaming has traditionally emerged especially in racist and sexist forms. It is also about holding individuals responsible for their own personal distress or difficulties instead of attributing responsibility to the transgressors who caused it.

 History of the concept

The phrase “Blaming the victim” was coined by William Ryan in his 1976 classic book of the same title, as a critique of Daniel Patrick Moynihan’s 1965 work The Negro Family: The Case for National Action, usually simply referred to as the Moynihan Report. Moynihan’s book summarized his theories about ghetto formation and intergenerational poverty. Ryan’s critique cast the Moynihan theories as subtle (and not so subtle) attempts to divert responsibility for poverty from social structural factors to the behaviors and cultural patterns of the poor. The phrase was quickly adopted by advocates for crime victims, in particular rape victims accused of abetting their victimization, although this usage is conceptually distinct from the sociological critique developed by Ryan.

Main article: Just-world phenomenon

It has been proposed that one cause of victim-blaming is the “just-world phenomenon”. People who believe that the world has to be fair may find it hard or impossible to accept a situation in which a person is unfairly and badly hurt. This leads to a sense that, somehow, the victim must have surely done ‘something’ to deserve their fate. Another theory entails the need to protect one’s own sense of invulnerability. This inspires people to believe that rape only happens to those who deserve or provoke the assault (Schneider et al., 1994). This is a way of feeling safer. If the potential victim avoids the behaviors of the past victims, then they themselves will remain safe and feel less vulnerable. A global survey of attitudes toward sexual violence by the Global Forum for Health Research shows that victim-blaming concepts are at least partially accepted in many countries. In some countries, victim-blaming is more common, and women who have been raped are sometimes deemed to have behaved improperly. Often, these are countries where there is a significant social divide between the freedoms and status afforded to men and women.

This idea dates from ancient times: the biblical Book of Job offers a refutation of the Just World Hypothesis, in which the main character, Job, maintains his faith through calamity after calamity, all of which are explicitly unrelated to his behavior, which remains devout.

Supporters of this view (once referred to as “Job’s comforters”) must perforce accept that to do otherwise would require them to give up their belief in a just world, and require them to believe in a world where bad things – such as poverty, rape, starvation, and murder – can happen to good men and women. Though a form of attribution error, this incorrect attribution differs from the “Fundamental Attribution Error” principally in its focus. Both concepts however center around a tendency to ignore situational contributors in favor of supposed internal failings on part of the subject being judged. In the Just-World Hypothesis the subject’s actions are not being scrutinized, but their situation; whereas those making the Fundamental Attribution Error tend to focus primarily on attributing actions to personal qualities and ignoring situational causes. Crimes or other events that create a victim give opportunity for both attribution errors – in blaming the victim for “allowing” themselves to be victimized by crime as well as the inability to cope afterwards. Despite their frequent simultaneity though, they remain two distinct attribution errors.

 Secondary victimization

Rape is especially stigmatizing in cultures with strong customs and taboos regarding sex and sexuality. For example, a rape victim (especially one who was previously a virgin) may be viewed by society as being “damaged.” Victims in these cultures may suffer isolation, be disowned by friends and family, be prohibited from marrying, be divorced if already married, or even killed. This phenomenon is known as secondary victimization.

Secondary victimization is the re-traumatization of the sexual assault, abuse, or rape victim through the responses of individuals and institutions. Types of secondary victimization include victim blaming and inappropriate post-assault behavior or language by medical personnel or other organizations with which the victim has contact. Secondary victimization is especially common in cases of drug-facilitated, acquaintance, military sexual trauma and statutory rape.

 Rape Shield Laws

In the United States, rape is unique in that it is the only crime in which there are statutory protections designed in favor of the accuser (known as “rape shield laws”). These were enacted in response to the common defense tactic of “putting the accuser on trial”. Typical rape shield laws prohibit cross-examination of the accuser (alleged victim) with respect to certain issues, such as his or her prior sexual history, or the manner in which he or she was dressed at the time of the rape. Most states and the federal rules, however, provide exceptions to the rape shield law where evidence of prior sexual history is used to provide an alternative explanation for physical evidence, where the defendant and the alleged victim had a prior consensual sexual relationship, and where exclusion of evidence would violate the defendant’s constitutional rights.