There's a lot of literature on this apparently, and I've only looked at one (55-page) 2018 NBER paper. There is some (correlational) evidence that the "pill mill" crackdown contributed to the switch to heroin, although this paper favors the oxy reformulation as the key event.
We attribute the recent quadrupling of heroin death rates to the August, 2010 reformulation of an oft-abused prescription opioid, OxyContin. The new abuse-deterrent formulation led many consumers to substitute to an inexpensive alternative, heroin. Using structural break techniques and variation in substitution risk, we find that opioid consumption stops rising in August, 2010, heroin deaths begin climbing the following month, and growth in heroin deaths was greater in areas with greater pre-reformulation access to heroin and opioids. The reformulation did not generate a reduction in combined heroin and opioid mortality—each prevented opioid death was replaced with a heroin death.
Regarding pricing the say:
Coplan et al. (2016) note that although the formulation for OxyContin changed, its price did
not. We find evidence consistent with this claim in the Truven Marketscan Research Database
(Marketscan). This is a database of individual-level claims for inpatient, outpatient, and prescription
drug use that by the end of our sample period provided information for over 37 million covered clients
per month from 350 self-insured plans.18 Figure 4 reports the monthly time series of the total price
and the price that patients pay out-of-pocket for oxycodone for 2006 through 2013. There is no large
change in either price series at the time of the reformulation and so it is unlikely that changes in the
legal price for oxycodone are driving substitution to heroin.
But I don't see discussion of the black market prices for oxy, only for heroin, which is indeed acknowledged as low.
They also reject some alternative hypotheses, including monitoring act(s):
A potentially important change in recent years has been the adoption of state-level PDMPs [ prescription drug monitoring programs ],
which are databases of prescriptions that doctors have written for patients. By giving doctors,
pharmacists, and in some cases law enforcement officials, access to this information, patients might
have greater difficulty obtaining large amounts of prescription drugs that can be abused and doctors
might be more conscious of their prescribing. A large body of research has studied the impacts of
PDMPs on prescribing and come to mixed results. While some find that PDMPs reduce opioid
overdose deaths (Kilby, 2015), others find no effects on prescribing patterns or effects for a very
limited subset of PDMPs (Buchmueller and Carey, 2018). Figure 10 shows the heroin death rate
separately for states that had passed PDMPs prior to 2010 and those that passed a PDMP in 2010 or
later. Death rates for states with a PDMP before 2010 and states with a PDMP in 2010 or later have
extremely similar heroin death rates over time. This suggests that PDMPs are unlikely to be causing
the abrupt rise in heroin death rates at the end of 2010. In addition, states began passing PDMPs in
2004 and have continued fairly steadily since then (National Alliance for Model State Drug Laws, 2014). One was created in 2004, two in 2005, two in 2006, four in each of 2007, 2008, and 2009, two
in 2010, four in 2011, and so on. Although the timing does not rule out the possibility that the PDMPs
impacted opioid prescribing and heroin deaths, it does strongly suggest that the PDMPs are not
responsible for the sharp, nationwide increase in heroin deaths that began at the end of 2010.
So they reject this monitoring issue in favor of their alternative (reformulation). Given that they say the body of research on monitoring is large, it's be worth looking for reviews of that for more detail.
And regarding pill mills... they have lot to say:
Beginning in 2009, a series of Federal and state programs were started that were designed to
reduce the impact of Florida’s pill mills. A number of authors have documented with a variety of
methods that the negative outcomes associated with opioids in Florida began to decline after the
introduction of these efforts (Johnson et al., 2014; Delcher et al., 2015; Rutkow et al., 2015; Chang
et al., 2016; Kennedy-Hendricks et al., 2016; and Meinhofer, 2016). If the Florida pill mills were a
significant component of OxyContin supply throughout the country, then the crackdown could also
be responsible for the shift to heroin in a way similar to the reformulation of OxyContin.
We investigate the pill mill hypothesis and find mixed evidence. We briefly summarize two
analyses that suggest the crackdown in Florida had little impact on the national increase in heroin
deaths and two that suggest it might have; the analyses can be found in Appendix B.
First, in Appendix Table B1, we provide a timeline of the significant events in the pill mill
crackdown in Florida. As the dates in the table suggest, the majority and potentially most effective
components of the pill mill crackdown did not go into effect until the second half of 2011, well after
the shift to heroin occurred. Second, we graph the time series of oxycodone and the seven other
opioids available in the ARCOS data for Florida and all other states. There does not appear to have been a reduction in any opioid in Florida starting in the third quarter of 2010 except for oxycodone
(see Appendix Figures B1a – B1h). In fact, there appears to have been slight increases in the use of
other opioids in Florida starting at that time. If the pill mill crackdown had been effective, then there
should likely have been reductions in all opioids that were being abused, not just oxycodone.
We do however find some evidence that states that were more exposed to the Florida pill
mills, and thus are more likely to be affected by the crackdown, see differential changes in the
growth of their heroin death rates. Our primary approach is based on anecdotal evidence from The
OxyContin Express which suggests that individuals who traveled to Florida to obtain opioids for
distribution in their home states were also using opioids. Using the universe of emergency
department and hospital admissions in Florida from 2007 through the second quarter of 2010, for
each state of residence, we calculate the admissions per capita for people aged 18-64 in Florida due
to opioids (labeled as OPCs), the non-opioid per capita admissions for the same group (NOPCs),
and generate the ratio, OPCs/NOPCs. We then designate states in the highest third of the
distribution as being more exposed to Florida’s pill mills. Our procedure identifies all states served
by The OxyContin Express, five states contiguous to these states (Alabama, Indiana, North
Carolina, West Virginia, Pennsylvania), and six other states (Rhode Island, Maine, New Jersey,
Maryland, Mississippi and New York) as likely affected by Florida’s pill mills. It is worth noting that
the procedure suggests that no states west of the Mississippi are being served by Florida’s pill mills.
In Figure 12, we graph the monthly heroin mortality for the states that are likely users of
Florida pill mills (black line) and all other states (grey line). The time trend for both series is very
similar prior to reformulation and both show a large change in slope starting near the August 2010
period. The increase in slope in the non-pill mill using states must be generated by some other
factor – a factor common to both sets of states. Fitting our quadratic spline through the monthly
data for the states unlikely to be pill mill users, the data suggests that the trend break occurs in
August of 2010. There is a noticeable break in trend for the pill mill states at the same period but
the trend break analysis suggests that the trend break occurs in October of 2011 – the month that all
components of the Florida pill mill crackdown law go into effect. This graph suggests that the
Florida reforms did not generate the initial shift to heroin but provides some evidence that the pill
mill crackdown in Florida also encouraged a shift to heroin.
[...] There is suggestive but not statistically significant evidence that the
pill mill crackdown in Florida appears to encourage more of a shift to heroin but only after October
2011 when the full set of reforms in Florida are in effect. That said, even in states that appear to have
little access to Florida pill mills, heroin mortality increased by a factor of 3.5 between August 2010
and the end of 2014, compared to a factor of 4.5 in pill mill access states. This indicates that at most,
the pill mill crackdown can explain 25 percent of the increase in heroin death rates in pill mill access
states between reformulation and the end of 2014.
I haven't bothered uploading any figures because SE/imgur is broken.