Article

Impact of prescription drug-monitoring program on controlled substance prescribing in the ED

a b s t r a c t

Objective: In 2009, Florida initiated a statewide prescription drug-monitoring program (PDMP) to encourage safer prescribing of controlled substances and reduce drug abuse and diversion. Data supporting the utility of such programs in the emergency department (ED) is scarce. This study sought to determine the effect of PDMP data on controlled substance Prescribing from the ED.

Methods: In this pre-post study utilizing a Historical control, pharmacists in the ED providED prescribers with a summary of the PDMP data for their patients. The number of controlled substances prescribed in the intervention group was compared with that prescribed in the historical control to determine if the intervention resulted in a change in the average number of controlled substance prescribed.

Results: Among the 710 patients evaluated, providing prescribers with PDMP data did not alter the average num- ber of controlled substance per patient prescribed (0.23 controlled substances per patient in the historical control compared with 0.28 controlled substances per patient in the intervention group; 95% confidence interval [CI],

-0.016 to 0.116; P = .125). All prescribers surveyed indicated that having PDMP data altered their controlled substance prescribing and felt more comfortable prescribing controlled substances.

Conclusions: Although the results did not demonstrate a change in the average number of controlled substances pre- scribed when prescribers were provided with PDMP data, results from the survey indicate that prescribers felt the data altered their prescribing of controlled substances, and thus were more contented prescribing controlled substances.

(C) 2015

Introduction

Nationally, emphasis on management of pain is reflected with in- creased rates of opioid prescribing. Between 2001 and 2010, the per- centage of emergency department (ED) visits where opioid analgesics were prescribed increased from 20.8% to 31.0%, a relative increase of 49% [1]. Pain continues to be a major impetus for patients who seek care in the ED, with 42% of all ED visit being related to pain, so it is no surprise that emergency medicine continues to rank among the top 5 specialties for opioid prescribing, contributing up to 39% of all narcotic

? Meetings: The results of this study were presented at the Florida Residency Con- ference in Gainesville, Florida on May 9, 2014.

?? Grants: There was no financial support provided for this study.

? Conflicts of interest: The authors have no conflict of interest.

?? Author contributions: MWM, PA, JS, and MS conceived the study, designed the tri-

al, and assisted with data collection. GB and KG also assisted with study design. DK and CS assisted with data analysis. MWM drafted the article, and all authors contributed substan- tially to its revision. MWM takes responsibility for the paper as a whole.

* Corresponding author at: Columbus Regional Health, Department of Pharmacy, 710 Center Street, Columbus, GA 31901. Tel.: +1 706 571 1495.

E-mail address: [email protected] (M.W. McAllister).

analgesics prescribed from the ambulatory setting [2-5]. The challenge for providers is to balance the undertreatment of pain with concerns of drug abuse and diversion [5].

Recently, several states have established guidelines for prescribing opioids from the ED to reduce Inappropriate prescribing. Some of the recommendations include prescribing small quantities of short-acting opioids for patients with acute pain and discouraging opioids for pa- tients with chronic pain. These guidelines also suggest using statewide prescription drug-monitoring programs (PDMPs) to review controlled substance prescription history before prescribing narcotics [6,7]. In ad- dition, the American College of Emergency Physicians released a clinical policy on opioid prescribing in the ED that mirrors these statewide guidelines. The guidelines discuss an intuitive benefit that PDMPs could provide, although they note the relative lack of data supporting their use, lack of standardization, and lack of interstate communication, which limit their utility [5].

In 2009, Florida’s legislature initiated a statewide PDMP titled The Electronic-Florida Online Reporting of Controlled Substances Evaluation program (E-FORCSE). E-FORCSE became operational on September 1, 2011, with the goals of encouraging safer prescribing of controlled sub- stances and reducing drug abuse and diversion within the state of

http://dx.doi.org/10.1016/j.ajem.2015.03.036

0735-6757/(C) 2015

Florida. E-FORCSE collects and stores prescribing and dispensing data for all controlled substances in schedules II, III, and IV dispensed within the state. Florida law requires that dispensing information be uploaded to the PDMP database within 7 days of dispensing a controlled substance prescription. Prescribers and pharmacists licensed within the state are able to register with E-FORCSE to access the on-line database [8].

Goals of this investigation

Given the paucity of data demonstrating a benefit of PDMPs in the ED, the primary end point of this study was to determine if providing prescribers E-FORCSE data alters the number of controlled substances prescribed. Secondary goals of the study were to determine if providing prescribers with E-FORCSE data affected the amount of opioid analge- sics prescribed, determine if the patient’s chief complaint or initial pain score affected the number of controlled substances prescribed, and survey prescribers in regards to their perceptions of the PDMP and their views on this project and its impact. This study also served as an opportunity to develop an efficient process for pharmacists prac- ticing in the ED to provide prescribers with E-FORCSE data in a way that could be easily incorporated into the pharmacists’ workflow.

Methods

Study design and setting

This was a pre-post study utilizing historical controls conducted in the ED of a tertiary care, urban university teaching hospital with a census of more than 95 000 adult ED visits per year. This study was specifically con- ducted in two intermediate care areas of the main ED. These areas typical- ly treat patients presenting with lower acuity chief complaints, many of which are discharged home after receiving treatment in the ED. The inter- vention was performed over a 2-week period in February 2014. For com- parison, patients in the historical control were seen and treated in the intermediate care areas over a 2-week period in December 2013. This study was approved by the institution’s investigational review board.

Selection of participants

Patients treated in the intermediate care areas of the ED were eligi- ble for the study if they were 18 years or older. Patients were excluded from analysis if they were not directly discharged from the ED (eg, ad- mitted to inpatient status, transferred to another facility, expired in the ED), had incomplete medical records, or if E-FORCSE data were not provided to the prescriber by an ED pharmacist. Informed consent was waived by the investigational review board for patients in this study as their information was collected retrospectively after the patient’s discharge from the ED.

Interventions

For the primary objective, three times a day during the intervention phase, pharmacists in the ED provided prescribers with EFORCSE data for all patients physically present in the intermediate care areas at that time. E-FORCSE data were evaluated and summarized by the pharmacist as well as whether the patient exhibited “drug-seeking behavior.” This re- port was prepared in a standardized progress note and included in the patient’s electronic medical record for review by the provider.

After the notes were completed, providers were notified via phone that this information was available for their review. For the intervention group, only those patients who had E-FORCSE information provided to the prescriber before patient discharge from the ED were included in the analysis.

For this study, drug-seeking behavior was defined as a patient with 4 or more controlled substance prescriptions of the same class (eg, opi- oids) written by providers from four or more different practice locations

within the most recent 12 months. This was based on the definition pre- viously used in the study by Weiner et al. [9]. For example, a patient that had received two prescriptions for hydrocodone/acetaminophen, 1 pre- scription for oxycodone, and 1 prescription for morphine all prescribed by different prescribers from different practice locations in 1 year would be considered to have exhibited drug-seeking behavior. For the second- ary objectives, patient data were retrospectively extracted from the electronic medical records as well as their E-FORCSE record.

Methods of measurement

For the historical control group, patient characteristics were collect- ed, including age, sex, chief complaint, and medications prescribed at discharge from the ED. Likewise, for the intervention group, patient de- mographics included age, sex, chief complaint, initial pain score (based on a 0-10 Likert scale), and medications prescribed from the ED. In ad- dition, data from E-FORCSE were extracted from the previous 12 months in the intervention group including number of controlled sub- stance prescriptions, number of physicians and pharmacies utilized, amount of opioids received, and determination of drug-seeking behav- ior using the aforementioned definition.

Outcome measures

The primary objective of this study was to determine the impact of PDMP data on the average number of controlled substance prescribed from the ED. The average number of controlled substances was defined as the total number of controlled substances prescribed per ED visit. Secondary objectives included comparison of opioid analgesic prescrib- ing between the historical and interventional groups, as well as the determination of possible associations was investigated between drug-seeking behavior and specific chief complaints along with drug- seeking behavior and the initial pain score. The PDMP data from pa- tients in the intervention phase were evaluated descriptively to deter- mine 12-month averages for number of controlled substance prescriptions received, amount of opioid medications received, and the number of prescribers and pharmacies utilized. Finally, prescribers in the ED were surveyed regarding their use of the E-FORCSE database, as well as their experiences and perceptions.

Statistical methods

An a priori sample size of 331 patients per group was determined with a two-sided type I error rate of 5%, and a power of 80% to detect a 10% difference in the average number of controlled substances pre- scribed per patient. Thus, the minimum target sample size was 331. To allow for a 5% dropout rate for exclusion criteria, our initial sample size goal was increased to 350 patients in each group. Baseline charac- teristics between groups were compared using a Wilcoxon rank sum test and a Pearson ?2 test when appropriate. For the primary outcome, comparison between controlled substance prescribing was performed with a Wilcoxon rank sum test. Comparisons between patients pre- scribed opioids were made with a Pearson ?2 test. Poisson regression models were used to control for the baseline characteristics of sex and age in the comparisons of chief complaint and intervention (historical control vs intervention group) for the number of controlled substance medications prescribed. Least square mean numbers of controlled sub- stance medications were estimated for significant effects and compared using the Tukey-Kramer adjustment for multiple comparisons. Associa- tions between drug-seeking behavior vs selected chief complaints and initial pain score were made using logistic regression models. To assess for interrater variation with respect to drug-seeking behavior, a re- searcher not involved with data collection evaluated 10% of the data for the primary end point, which was used to calculate a ? statistic. Since other variables (eg, number of controlled substances prescribed) are objective in nature, agreement was not assessed on these variables.

Table 1

Demographic and patient characteristics

Characteristic

Historical control (n = 354)

Intervention (n = 356)

P

Age, mean (SD)

44.19 (15.74)

44.31 (14.59)

.872a

18, 44, 88

19, 45, 88

Sex, male, no. (%)

156 (43.7%)

149 (41.8%)

.610b

Complaint

Abdominal pain (atraumatic, acute)

34 (9.6%)

46 (12.9%)

.0001b

Acute, atraumatic pain not otherwise specified

54 (15.3%)

64 (18.0%)

Chest pain (atraumatic, acute)

11 (3.1%)

38 (10.7%)

Chronic pain

33 (9.3%)

26 (7.3%)

Dental pain (atraumatic, acute)

15 (4.2%)

12 (3.4%)

Headache (atraumatic, acute)

19 (5.4%)

24 (6.7%)

Pain resulting from trauma

61 (17.2%)

65 (18.3%)

Nonpain complaint

127 (35.9)

81 (22.7%)

a Wilcoxon rank sum test.

b Pearson ?2 test.

All statistical analyses were run using SAS Version 9.3 for Windows. Sur- vey data were analyzed with simple counts using Microsoft Excel 2013.

Results

Characteristics of study subjects

A total of 804 patients were enrolled in the study. Ninety-eight pa- tients were eliminated from the intervention portion of the study be- cause they required admission to the hospital, with 356 patients remaining in the intervention group and 354 in the historical control group. Patient characteristics were similar between the 2 groups, except that the intervention group had more patients with painful complaints and more patients with chest pain as compared with the historical con- trol group (Table 1).

Main results

There was no change in the average number of controlled substances prescribed per patient when E-FORCSE data were provided to pre- scribers in the ED (0.28) compared to the historical control group (0.23) (95% confidence interval [CI], -0.016 to 0.116; P = .125). The analysis for the primary end point was reevaluated after removing pa- tients with a nonpainful complaint from both groups, and this also re- sulted in no statistical differences. A researcher not involved with data collection evaluated 10% of the data for the primary end point, which was used to calculate a ? statistic of 0.86 indicating high agreement be- tween observers (Table 2).

Poisson regression models were used to investigate if group (histor- ical control vs intervention group), sex, and type of complaint (eg, chest pain, dental pain, headache) affect the number of controlled substances prescribed from the ED. When only group (historical control vs inter- vention group) was considered in the model, it was found that the ex- pected number of controlled substance prescriptions was not different between groups (P = .166). Once sex and complaint were added to the model, the only significant predictor of number of controlled sub- stances prescribing was complaint (P b .0001). Least square means are summarized in Table 3, ranging from a low of 0.014 controlled sub- stance prescriptions per patient for a nonpain complaint to a high of

0.593 for dental pain. When the complaints are compared pairwise, each complaint other than chest pain (atraumatic and acute) differs sig- nificantly from nonpain complaint. Chest pain also differs significantly from dental pain and from pain resulting from trauma. No other pair of complaints differs significantly.

When the secondary outcomes were analyzed, there was no differ- ence overall in the percentage of patients who were prescribed opioids between the two groups (historical control 19.5% vs. intervention group 23.6%; P = .184).

Logistic regression models were used to determine if complaints and initial pain score affect the probability of drug-seeking behavior. Dental pain was excluded from this analysis because none of these patients ex- hibited drug-seeking behavior. First, models with only complaint and with only pain score as predictors (separate models) were fit. When only complaint was considered in the model, the complaint was a signif- icant predictor of drug-seeking behavior. Patients with chronic pain were almost 10 times more likely to have drug-seeking behavior com- pared with those with a nonpain complaint (odds ratio [OR], 9.58; 95% CI, 2.26-40.56; P = .0008). None of the other comparisons to nonpain complaint were significant. When only initial pain score was considered in the model, the initial pain score was a significant predictor of drug- seeking behavior. The odds of drug-seeking behavior increased by about 34% for each 1 unit increase in pain score (OR, 1.34; 95% CI, 1.10-1.64; P = .004). This means that the odds of drug-seeking behavior increased by a factor of 1.34 for each 1 unit increase in pain score. When both complaint and initial pain score were entered in the model, com- plaint was no longer significant (P = .169). The odds of drug-seeking behavior increased by about 30% for each 1 unit increase in pain score (OR, 1.30; 95% CI, 1.05-1.60; P = .014).

Providers practicing in the ED were surveyed as to their use of the

statewide PDMP (E-FORCSE). They were also asked if they felt that the in- tervention by the pharmacist altered the way they prescribed controlled substances; results are listed in Table 4. Of 25 respondents, only 9 (36%) had access to the PDMP system, and of those 9 providers with access, 6 (66%) “rarely” or “never” accessed the system. Providers listed many lim- itations to the use of the system. The most common issues were with login or access, as well as time required to utilize the system. All respon- dents felt that the data provided by the pharmacist were beneficial and that it altered their prescribing of controlled substances. All respondents

Table 2

Primary outcome

Outcomes

Historical control (n = 354)

Intervention group (n = 356)

P

No. of controlled

mean (SD)

0.28 (0.47)

0.23 (0.43)

.125a

substance Rx

min, median, max

0, 0, 2

0, 0, 2

No. of noncontrolled

mean (SD)

0.84 (1.05)

0.96 (1.03)

.080a

substance Rx

min, median, max

0, 1, 11

0, 1, 6

a Wilcoxon rank sum test.

Table 3

Least-square (adjusted) mean number of controlled substance prescriptions estimated using the Poisson regression model

prescribers in the state registered to use the system [8]. Current PDMPs, including the one in Florida, have many complex issues includ- ing issues with login or access, delays in data updates, lack of interstate

Complaint

Mean

LL CI mean

UL CI mean

communication, and time required for providers to retrieve the infor-

Dental pain (atraumatic, acute)

0.593

0.363

0.967

mation, which limit their utility, especially in an ED setting [5].

Pain resulting from trauma

0.524

0.412

0.667

This study set out to determine if providing emergency medicine

Chronic pain

0.390

0.259

0.587

providers with a summary of their patients’ prescription monitoring

data would alter controlled substance prescribing. In our study, we were unable to demonstrate a difference in controlled substance pre-

Chest pain (atraumatic, acute)

0.102

0.042

0.245

scribing or opioid prescribing when prescribers were provided with

Nonpain complaint

0.014

0.005

0.045

their patients’ E-FORCSE data compared to those who were not. We

Abdominal pain (atraumatic, acute)

0.288

0.191

0.433

Acute, atraumatic pain NOS

0.280

0.199

0.393

Headache (atraumatic, acute)

0.233

0.125

0.432

LL indicates lower level; UL, upper level.

also noted that they felt more comfortable prescribing controlled sub- stances when they had the data provided by the pharmacist.

Discussion

Pain is one of the most common complaints for many patients pre- senting to the ED and emergency medicine providers must balance inef- fective or undertreatment of pain with inappropriate prescribing, which can facilitate drug abuse and diversion [5]. Recent statewide guidelines on opioid prescribing in the ED, as well as guidelines from the American College of Emergency Physicians, mention the use of statewide PDMPs in the ED, although they note there are many limitations to their use and little data supporting their proposed benefits. One such study is that of Baehren et al., in which providers in the ED were provided with the PDMP data of their patients and evaluated if this had any im- pact on opioid prescribing. Of the 179 patients who completed the study, opioid prescribing was altered in 41% of the cases, with 61% of those patients receiving fewer or no opioids after physicians were shown the PDMP data [10].

Since the implementation of E-FORCSE, Florida has seen promising results, as demonstrated by a 41% decrease in oxycodone-related deaths from 2011 to 2012 and a 10% decrease in overall drug related deaths during the same period [11]. During this period, the state has also seen a decrease in the amount of “doctor shopping” as demonstrated by a 50.6% decrease in the number of individuals who had controlled substances prescribed to them by more than 5 prescribers and dis- pensed at more than 5 pharmacies in a 90-day period [8]. Unfortunately however, the program is still underutilized, with only 10% of all

evaluated a change in the number of controlled substance prescribing, expecting that when prescribers had access to their patients’ controlled substance history that they may prescribe controlled substances less frequently. However, as shown in the study by Baehren et al., when phy- sicians are provided with prescription monitoring data, not all patients received less opioids, some patients received more, possibly moving to- ward a “more appropriate prescribing” of opioids rather than a true de- crease in prescribing of opioids. Another possible explanation for the difference in findings between these studies is that, in contrast to the study by Baehren et al., providers in our study were not required to uti- lize the prescription monitoring data that was provided to them. Al- though our providers did feel more comfortable prescribing controlled substances when these data were provided to them, it was not assessed to what extent they were actually reviewing the E-FORCSE data provid- ed to them by the pharmacists, which may limit the interpretation of our results. In our study, we found that providers felt more comfortable prescribing controlled substances when they had the patient’s con- trolled substance history available to them. If providers prescribed more controlled substances (as seen in some of the patients in the study by Baehren et al.) because they felt comfortable that these pa- tients were not drug seeking, we may in fact be moving toward “more appropriate prescribing” of controlled substances, although a numerical difference is not apparent.

Twenty-eight patients in our study met our definition of drug-

seeking behavior. We found that when only considering chief com- plaint, drug-seeking behavior was associated with those patients pre- senting with chronic pain, although when initial pain score was taken into account, this association disappeared and drug-seeking behavior was more closely associated with the patients pain score. We discovered that, as pain scores increase, so does the chance that the patient exhibits drug-seeking behavior, by about 30% for each 1 unit increase in pain

Table 4

survey results

Survey question Results (%)

Providers with an E-FORCSE login (n = 25) 9 (36)

If you have a login, how often do you utilize E-FORCSE? (n = 9)

Always –

Often 3 (33)

Sometimes –

Rarely 5 (55)

Never 1 (11)

What limits your ability to utilize E-FORCSE? (n = 28)a

Do not have any issues with E-FORCSE

Do not have a login

10 (36)

Have problems with login/access

12 (43)

Time required to access patient data

6 (21)

Did you feel that the E-FORCSE data provided to you was beneficial? (n = 18) Very beneficial

12 (66)

Somewhat beneficial

6 (33)

Indifferent

Not beneficial

Felt that intervention altered controlled substance prescribing (n = 18)

18 (100)

Did you feel more or less comfortable prescribing controlled substances when you had E-FORCSE data provided to you? (n = 18)

More comfortable

18 (100)

Indifferent

Less comfortable

a Several prescribers selected multiple limitations.

score. Future studies might target patients with chronic pain or those with higher initial pain scores as they were associated with our defini- tion of drug-seeking behavior.

We were able to establish an effective and efficient means of provid- ing prescribers with E-FORCSE data using the hospital’s electronic med- ical record. Although it was not originally an outcome of the study, we randomly conducted time-in-motion studies, starting roughly half way through the project to determine how much time was involved in the process. It was determined that, with the current system in place, the pharmacist in the ED could access the E-FORCSE system, ex- tract and analyze the data, summarize the findings, and provide the data to the prescriber via the electronic medical record in approximately 4 minutes per patient.

Limitations

By attempting to develop a way to efficiently provide prescribers E- FORCE information amidst normal patent care functions, we were unable to assess whether prescribers intended to prescribe a controlled sub- stance before reviewing our E-FORCSE information, as was done in the study by Baehren et al. Because of this, we were unable to assess the num- ber of patients whom would have received a controlled substance if not for our intervention. In addition, we were unable to evaluate to what ex- tent providers may have utilized the E-FORCSE system during our histor- ical control, although we did note in our survey that very few of our providers were accessing this information on a regular basis.

Another limitation of our study was our definition of drug-seeking behavior. It is very difficult to make an objective definition of something as subjective as patient behavior. We attempted to use a previously pub- lished definition to allow for consistency is the literature, although there seems to be limitations with this definition. Several of our patients who met our criteria for drug-seeking behavior had previous diagnoses of chronic pain. By assessing only if patients had received prescriptions from physicians at multiple office locations, this may inappropriately classify patients as drug-seekers who are actually seeking care from sev- eral providers in a large practice group that happens to have multiple of- fice locations, which would not be an uncommon occurrence. Future studies should attempt to refine the definition of drug-seeking behavior. The premise of the study was that we attempted to find a change in the average number of controlled substances prescribed, although a pre- vious study demonstrated that by providing prescribers with prescription monitoring data, some patients were prescribed fewer opioids and some were prescribed more [10]. Because of this, the true impact of this study may possibly be increasing the appropriateness of controlled substance prescribing rather than a numerical change in the number of controlled

substances prescribed. Finally, our baseline characteristics were unequal in regards to painful complaint, which may have affected on our results.

Conclusions

Providing ED prescribers with E-FORCSE data did not alter the aver- age number of controlled substances prescribed per patient. In addition, providing such information did not alter opioid prescribing, although prescribers felt that these data altered the way they prescribed con- trolled substances. In addition, prescribers felt more confident prescrib- ing controlled substances when statewide PDMP data were provided to them by pharmacists practicing in the ED. In addition, pharmacists prac- ticing in the ED provided prescribers with E-FORCSE data in a way that could be incorporated into their daily workflow.

References

  1. Mazer-Amirshahi M, Mullins PM, Rasooly I, van den Anker J, Pines JM. Rising opioid prescribing in adult U.S. emergency department visits: 2001-2010. Acad Emerg Med 2014;21:236-43.
  2. Pletcher MJ, Kertesz SG, Kohn MA, Gonzales R. Trends in opioid prescribing by race/ ethnicity for patients seeking care in US emergency departments. JAMA 2008; 299(1):70-8.
  3. Volkow ND, McLellan TA, Cotto JH. Characteristics of Opioid prescriptions in 2009. JAMA 2011;305:1299-301.
  4. National Center for Health Statistics. Medication therapy in ambulatory medical care: United States, 2003-04. Vital and Health Statistics, Series 13, number 163; 2006[Avail- able at http://www.cdc.gov/nchs/data/series/sr_13/sr13_163.pdf#page=26#page=

    26. Accessed on May 20, 2014].

    Cantrill SV, Brown MD, Carlisle RJ, Delaney KA, Hays DP, Nelson LS, et al. Clinical pol- icy: critical issues in the prescribing of opioids for adult patients in the emergency department. Ann Emerg Med 2012;60:499-525.

  5. Washington Emergency Department Opioid Prescribing Guidelines. Available at http://washingtonacep.org/Postings/edopioidabuseguidelinesfinal.pdf; 2011.

    [Accessed on May 20, 2014].

    New York City Emergency Department Discharge Opioid Prescribing Guidelines Clinical Advisory Group. New York City Emergency Department Discharge Opioid Prescribing Guidelines. New York City website http:// www.nyc.gov/html/doh/downloads/pdf/bases/opioid-prescribing-guide- lines.pdf. [Accessed May 20, 2014].

  6. 2012-2013 Prescription Drug Monitoring Program Annual Report. Florida Department of Health; 2013[Available at http://www.floridahealth.gov/statistics-and-data/e-forcse/ news-reports/_documents/2012-2013pdmp-annual-report.pdf. Accessed March 31, 2014].
  7. Weiner SG, Griggs CA, Mitchell PM, Langlois BK, Friedman FD, Moore RL, et al. Clini- cian impression versus prescription drug monitoring program criteria in the assess- ment of drug-seeking behavior in the emergency department. Ann Emerg Med 2013;62:281-9.
  8. Baehren DF, Marco CA, Droz DE, Sinha S, Callan EM, Akpunonu P. A statewide pre- scription drug monitoring program affects emergency department prescribing be- haviors. Ann Emerg Med 2010;56:19-23.
  9. Drugs identified in deceased persons by Florida Medical Examiners 2012 Report. http:// www.fdle.state.fl.us/Content/getdoc/0f1f79c0-d251-4904-97c0-2c6fd4cb3c9f/MEC- Publications-and-Forms.aspx; 2013. [Accessed on March 31, 2014].

Leave a Reply

Your email address will not be published. Required fields are marked *