Efficiency and Effectiveness of Using Nonmedical Staff During an Urgent Mass Prophylaxis Response

Abstract

Using a simulated anthrax scenario, the Philadelphia Department of Public Health tested the readiness of a nonmedical closed point-of-dispensing (POD) site to see how rapidly and accurately it could provide medication to its internal population. This closed POD had developed and exercised its mass prophylaxis plan in conjunction with the local health department twice before, and the department was interested in assessing the impact of having no onsite department involvement. Two sessions were conducted as part of the overall exercise. In session 1, agency staff ran POD operations with no department involvement. During session 2, department staff provided an hour-long training session and oversaw POD operations. Mean throughput and accuracy rates of the 2 sessions were then compared to a previous health department public POD exercise staffed by department personnel and medical volunteers. The closed POD would be able to process the entire internal population in an estimated mean time of 23.9 hours. The accuracy rates for dispensing the correct medication during session 1 was 84.7% and 92.4% during session 2 (p=0.0012). Overall accuracy was significantly higher in a previous local health department public POD exercise (88.6% vs. 96.9%, p < 0.0001), as was pediatric dosing accuracy (p < 0.0001). We concluded that nonmedical closed PODs are a valuable strategy during a public health emergency that requires large segments of a population to receive medication rapidly. They must be activated judiciously, however, as their use may increase adverse events and potentially result in discontinuation of antibiotic prophylaxis should people choose not to finish the course. Local health department training and oversight reduce errors but may not always be available.

The 2001 anthrax attacks in the United States resulted in 22 cases; 11 were of the inhalational and more lethal variety, and 5 resulted in fatalities.^1-5 More than 10,000 people in 5 states were potentially exposed. At the time, the Centers for Disease Control and Prevention (CDC) recommended empirical prophylaxis of 60 days of oral doxycycline or ciprofloxacin for all potentially exposed people, including pregnant women and children.⁶ Prophylaxis includes antibiotics such as doxycycline, ciprofloxacin, and amoxicillin, but they must be given within the first few days of exposure for optimal prevention.^4,6 Even with supportive care in a hospital, approximately 50% to 75% of symptomatic inhalational anthrax cases are fatal.⁷ Rapid dispensing of mass prophylaxis is crucial to prevent mass casualties and overburdening of the healthcare system.^3,5 Current recommendations for postexposure prophylaxis for inhalational anthrax are 60 days of oral antibiotics with concomitant administration of a 3-dose series of anthrax vaccine.⁸

For a large-scale bioterrorist attack of anthrax spores, public health departments would rely on the mass quantities of doxycycline and ciprofloxacin available in the Strategic National Stockpile (SNS) to rapidly medicate exposed people. The SNS provides an initial 10-day supply of medication in the first shipment, with the additional 50-day supply arriving later. Under routine medical care, these antibiotics are contraindicated in select populations (eg, those with severe allergies, pregnant women, pediatric populations, and those taking certain medications for chronic illnesses) and would be dispensed under an emergency use authorization. To expedite the screening process, local health departments have reduced the number of questions asked to only a few critical ones that focus on severe adverse reactions and maternal and child health considerations. Understanding the safety and risks with both emergency and long-term use of these antibiotics is essential for public health officials when deciding to activate an emergency mass prophylaxis response.

Tetracycline and quinolone antibiotics, such as doxycycline and ciprofloxacin, are sometimes poorly tolerated during routine treatment. Martin et al reported that 28% of participants in a prospective cohort study comprised of those potentially exposed after the 2001 anthrax attacks reported at least 1 adverse event, and 14% self-reported a moderate or severe adverse event while taking these medications.⁴ Common side effects of ciprofloxacin and doxycycline are nausea, diarrhea, vomiting, abdominal pain, headache, and rash, often leading the patient to discontinue their use.⁴ The list of potential adverse events for ciprofloxacin is considerably longer than doxycycline and includes (but is not limited to) increased susceptibility to seizures, gastrointestinal symptoms, myasthenic exacerbation, and drug interactions with those on probenecid, theophylline, glyburide, cyclosporine, and warfarin.⁹ Doxycycline is listed as Pregnancy Category D by the Food and Drug Administration; evidence exists demonstrating teratogenic effects in humans. Ciprofloxacin is listed as Category C; human teratogenic evidence is lacking but theoretically could occur as suggested by evidence in animal studies.¹⁰ Amoxicillin typically has fewer complications when compared to doxycycline and ciprofloxacin and is routinely used for pregnant and pediatric populations.¹¹ A retrospective cohort analysis by Meropol et al found long-term use of doxycycline, ciprofloxacin, and amoxicillin to be generally safe but documented increases in adverse events associated with ventricular arrhythmias, renal toxicity, hepatic toxicity, and pseudomembranous colitis for those on ciprofloxacin compared to doxycycline and amoxicillin.¹²

More important, antibiotics taken systemically are sometimes associated with adverse reactions, some so severe they require medical intervention. Shehab et al estimate 142,505 emergency department visits occur in the United States annually because of adverse events directly related to use of antibiotics (95% confidence interval [CI], 116,506-168,504).¹³ Approximately 79% of these emergency department visits were attributed to severe allergy, defined as an immunological reaction consisting of rash or anaphylaxis. Doxycycline and ciprofloxacin contributed 2.3% and 3.5% of all emergency department visits, respectively, but the penicillin class, which includes amoxicillin, caused the most emergency department visits (37%).¹³

In response to the 2001 attacks, the CDC started the Cities Readiness Initiative to address potential anthrax dissemination in large urban areas and improve the capacity of local and state jurisdictions to deliver medications during a large-scale public health emergency.^7,14,15 Local health departments take the lead in planning to ensure that all potentially exposed people in their communities receive medication within 48 hours after the decision is made to recommend prophylaxis.^16,17 Given the urgency, magnitude of response, and time constraints, strategies to dispense oral medications are some of the most logistically challenging emergency response operations public health agencies face. One way to distribute these medications is through the establishment of public points of dispensing (PODs).^7,18-20 POD staff typically come from a medical background and are drawn from health departments or volunteer organizations, such as the Medical Reserve Corps (MRC).¹⁵ But expected staff absenteeism and lack of sufficient medical personnel make staffing PODs very challenging during a public health emergency. To reduce the number of PODs and personnel needed for a response, some local health departments use the head-of-household model, in which only 1 person per household reports to a POD and picks up medications for the entire family. To further decrease demand for POD staff, many jurisdictions have developed a closed POD model and collaborated with healthcare agencies, private businesses, and government agencies to allow their nonmedical personnel to screen for and dispense medications to their internal populations.⁷

Literature describing mass prophylaxis exercises with nonmedical dispensing models is sparse and typically focuses on time-flow and processing rates.^15,17,21-23 However, accuracy is another key component in prophylaxis dispensing.²³ Few articles focus on the consequences of medication errors and adverse events when nonmedical staff are used. Ablah et al examined medication errors among PODs that used nonmedical staff and found that, although public PODs and first responder facilities had higher throughput rates, hospital facility PODs had a lower incidence of medication errors.²⁴ A study by Spitzer et al used a predominantly nonmedical POD but studied only throughput rates.²² The use of nonmedical staff to dispense certain antibiotics potentially increases error rates and, thus, may also increase adverse events with respect to allergies, fetal abnormalities, other drug interactions, and incorrect dosing for children.

A previous emergency simulation exercise using traditional staff resources, such as health department staff and trained MRC volunteers, observed that error rates range from 3% to 4% with respect to accurately dispensing medications to the general population.²³ The authors postulated that an absence of health professionals among closed POD staff would increase error rates in dispensing medication. This could lead to an increase in adverse events, potentially making such closed POD activation detrimental. Moreover, public health departments recognize that their key response staff will be extremely busy. They may not be able to go to closed PODs to coordinate set-up or train staff. Nonmedical closed PODs present additional challenges as public health departments depend on outside leadership (often without medical training) to oversee and run a medical countermeasure operation. Therefore, health departments depend on previous training sessions and just-in-time training to fully support closed POD operations. This quantitative analysis of dispensing errors and associated adverse events during a nonmedical closed POD exercise provides insight as to the consequences and implications of using this response strategy.

In 2012, a nonmedical government agency, in conjunction with the Philadelphia Department of Public Health, tested its closed POD plan to provide medications to employees and their household members (a total population of 23,200 people). This nonmedical government agency developed its closed POD plan in 2009 and had participated in 2 training sessions and exercises in which staff viewed health department–led presentations and then performed a pill-dispensing exercise. The local health department assessed how quickly and accurately closed POD staff would dispense medicines to their population and then compared these rates to a previous local health department public POD exercise that used department and medical staff. The health department also measured the impact that department-led training and supervision would have on efficiency compared to a closed POD activating and running on its own. Estimations of throughput and expected adverse events can inform public health officials when they are deciding whether to recommend mass prophylaxis and activation of closed PODs in emergencies and can reveal training needs for closed PODs. The questions posited by the local health department for this activity were:

1. Can people who have been previously trained in a closed POD activate and dispense medication without onsite support from the local health department in the allotted timeframe?

2. What are the error rates for dispensing contraindicated medications and pediatric dosages?

3. What is the impact on efficiency and accuracy if the local health department provides onsite training and supervision of closed POD activities?

Methods

Exercise and Training

Two sessions were conducted as part of the overall exercise simulating closed POD activation in response to an anthrax exposure. In session 1, government agency leaders followed their closed POD plan, provided just-in-time training to closed POD staff, and oversaw closed POD operations without the involvement of local health department staff. Just-in-time training included a review of POD roles, step-by-step instructions, and forms with specific role details, which include screenshots of the head-of-household form, screening algorithm, and pediatric dosing guidelines. Job action sheets giving details of each station's responsibilities were also provided. Agency staff not assigned a specific role played the part of patients going through the POD.

During session 2, local health department staff provided an hour-long training session, including presentations describing closed POD operations and staff roles. A video showing POD set-up also was shown. Health department staff acted as POD leadership staff, provided the just-in-time training to closed POD staff, and oversaw the operation of the closed POD. Different staff members from the government agency were chosen to act as closed POD staff to better evaluate staff performance and provide employees more opportunities to practice being closed POD staff. For this specific exercise, set-up of the POD stations and distribution of materials were conducted by the local health department staff prior to both sessions (Table 1).

Table 1.

Comparison of Methods Between 2 Sessions of a Nonmedical Closed POD Exercise

Activity	Session 1	Session 2
1. POD set-up	Conducted by local health department staff	Conducted by local health department staff
2. Closed POD role assignments	Local health department staff pre-assigned leadership roles	Local health department staff pre-assigned leadership roles
3. Training materials	Provided to all participants at registration	Provided to all participants at registration
4. Orientation	None provided	2-hour department-led training
5. 15-minute just-in-time training	Provided by closed POD leaders	Provided by health department staff
6. Patients went through POD and were processed	Played by closed POD staff	Played by closed POD staff

Patients and Head-of-Household Forms

Eighty different sets of head-of-household forms and scripts were developed in advance. These forms simulated average household size, family composition, and certain medical histories (eg, rates of non-severe drug side effects, severe drug allergies, and pregnancy) in Philadelphia. A key was developed that listed the correct medication and dosage assignment for every individual listed on a head-of-household form. Screening algorithms used doxycycline as the first-choice antibiotic, with ciprofloxacin and amoxicillin as the second and third choices. Adults with no contraindications were sent to express dispensing stations and automatically received doxycycline. The screening algorithm indicated that pregnant women and those with severe allergies to tetracyclines (defined as rash, hives, swelling, or respiratory distress), including doxycycline, should receive ciprofloxacin. Anyone with severe allergies to both doxycycline and ciprofloxacin would receive amoxicillin. Anyone with a head-of-household form that included children went through screening, not express dispensing. Children received doxycycline if they did not have a severe preexisting allergy to tetracyclines, and all children required a dosing adjustment if under 9 years of age.

During the closed POD exercise sessions, patients picked up the completed head-of-household forms and scripts at greeting stations, and an entrance time was recorded on the back of the form. Patients were directed to either express dispensing or screening, depending on the content of their forms. Patients directed to screening stations were questioned by screening staff and assigned a medication type and dose based on the answers provided by the head of household. Using the medication and dose assignments made by screening staff on the head-of-household form, dispensers provided individuals with the appropriate number and type of medication bottles, as well as the appropriate drug and disease information sheets. Dispensers simulated distribution of the medication by writing the medication and dose assignment given to each patient on stickers that were then placed on the head-of-household forms. Dispensers were responsible for ensuring that screening staff had made a medication assignment for each individual listed on the form. Patients going through express dispensing were automatically assigned the adult dose of doxycycline. Upon exiting the POD, patients turned in the head-of-household forms, and time of exit was recorded on the back of the form. Screener and dispenser medication assignments were compared to the key to determine the accuracy of medication and dose designation.

Analysis

Estimated throughput of each session was calculated using the number of people included on the head-of-household forms processed per session, divided by time period. The number of people per household was based on the average number of people in a household for the local jurisdiction. Head-of-household forms without entry and exit timestamps were excluded from analysis.

Screener and dispenser medication assignments were compared to the medication assignment key, and accuracy rates were calculated for both sessions. Analysis of variance tests were performed using SAS software, Version 9.3 (SAS Institute, Inc., Cary, NC) to compare the accuracy of the type and dose of medication prescribed between the 2 sessions. Medication assignment accuracy was also assessed for those with severe medication allergies and pregnant women. Dispenser dosing error for children (those under 9 years old) was categorized based on whether a child was dosed correctly or not, and then further categorized by incorrect dosing directionality. The difference between the medication dose assigned to children at dispensing and the correct medication dose in the medication assignment key was compared between the 2 closed POD sessions using a t-test. Screening forms with incomplete or missing information were excluded from the pediatric dosing assignment analysis. Accuracy data collected from this closed POD exercise were then combined and compared to a 2005 public POD dispensing exercise run by the same local health department, using medically licensed professionals.²³

Results

Throughput Rate

A total of 151 people affiliated with the government closed POD participated as either patients or staff in both sessions of the exercise. During the first session, 165 head-of-household forms were processed (estimated throughput of 914.4 people medicated/hour), which would allow for the provision of medications to the entire closed POD population of 23,200 individuals in 25.4 hours (given that it was continuously operated). During session 2, 192 head-of-household forms were processed (estimated throughput of 1,040 people medicated/hour), which would allow for the provision of medications to the entire closed POD population of 23,200 individuals in 22.3 hours (Table 2). The combined estimated time needed to medicate the entire closed POD population is 23.9 hours.

Table 2.

Estimated Throughput Rates and Dispensing Times for a Nonmedical Closed POD

Session	Head-of-Household Forms Processed (N)	People Processed Contained in Form	Session Time (minutes)	Estimated Throughput (heads of household/hour)	Estimated Throughput (people/hour) ^a	Estimated Time Needed to Medicate Entire Dispensing Population (hours) ^b
Session 1	165	381	25	396	914.4	25.4
Session 2	192	364	21	549	1,040.0	22.3
Combined	357	745	46	466	971.7	23.9

The number of people was derived from data on head-of-household forms, which was based on the assumption of 2.45 people medicated per head of household.²⁵

The closed POD dispensing population is 23,200 individuals, including staff and staff household members.

Dispenser Accuracy Rates

In session 1, we observed a 9.3% error rate in dispensing the correct type of antibiotic and a 15.8% error rate in dispensing to those with a severe allergy to tetracyclines. Session 2 saw a significant decrease in error rates for dispensing the correct type of medication (4.4%, p=0.01). The error rate for correctly dispensing the right type of medication to those with a severe allergy to tetracyclines also decreased to 7.7% in session 2, although this decrease was not statistically significant (p=0.1926) (Table 3).

Table 3.

Nonmedical Closed POD Dispensing Accuracy (Session 1 vs. Session 2)

Group	Total (N)	Number (%) with Complete Dosing Information ^c	Number (%) Accurately Prescribed	Unadjusted Chi-Square P Value^d
All household members (both type and dose of medication)
2012 Exercise (both sessions)	745	696 (93.4%)	617 (88.6%)	0.0012
Session 1	381	339 (89.0%)	287 (84.7%)
Session 2	364	357 (98.1%)	330 (92.4%)
All household members (type of medication only)
2012 Exercise (both sessions)	745	718 (96.4%)	669 (93.2%)	0.01
Session 1	381	356 (93.4%)	323 (90.7%)
Session 2	364	362 (99.5%)	346 (95.6%)
All household members (dose of medication only)
2012 Exercise (both sessions)	745	696 (93.4%)	658 (94.5%)	0.0046
Session 1	381	339 (89.0%)	312 (92.0%)
Session 2	364	357 (98.1%)	346 (96.9%)
Severe medication allergies^a
2012 Exercise (both sessions)	109	109 (100.0%)	96 (88.1%)	0.1926
Session 1	57	57 (100.0%)	48 (84.2%)
Session 2	52	52 (100.0%)	48 (92.3%)
Pregnant women^b
2012 Exercise (both sessions)	52	52 (100.0%)	49 (94.2%)	0.6495
Session 1	24	24 (100.0%)	23 (95.8%)
Session 2	28	28 (100.0%)	26 (92.9%)

Severe allergies include rash, hives, impaired breathing, and swelling, and those with unknown allergy symptoms.

Considered inaccurate when individual was prescribed doxycycline.

Complete dosing information to judge accuracy includes drug name, dose, and frequency of administration.

Chi-square value for comparison between session 1 and session 2 of the 2012 closed POD exercise.

During the first session, 4.2% of pregnant women were erroneously dispensed doxycycline. The error rate for dispensing doxycycline to pregnant women in session 2 was 7.1%. This increase in error rate was not statistically significant (p=0.6495). Dispensers provided the incorrect dose for 8.0% of the total household members (n=27/339) during session 1 and 3.1% of the total household members (n=11/357) during session 2 (p=0.0046). Approximately half (52.6%, n=20/38) of these errors occurred for patients under 9 years of age.

Pediatric Dosing Accuracy Rates

Of the 28 observations in which the dosing assignment was provided for children under 9 years of age, 8 (28.6%) were dosed accurately based on the child's reported age and weight. Of the 20 incorrect dosages assigned, 55% were overdosed and 45% were underdosed. Only 2 of 20 (10.0%) of the dosing assignment errors were within 0.5 teaspoons' difference from the correct dosage (data not shown).

Comparison to Traditional Medically Staffed PODs

Overall dispensing accuracy was significantly higher in the 2005 public POD exercise (96.9%) when compared to the 2012 closed POD exercise (88.6%, p<0.0001). The accuracy rate for assigning the correct type of medication to individuals with severe medication allergies in the 2005 public POD exercise (96.3%) was not found to be significantly higher than in the 2012 closed POD exercise (88.1%, p=0.0623). Pediatric dosing accuracy was significantly higher in the 2005 exercise (89.6%, n=181/202) than in the 2012 exercise (28.6%, n=8/28, p<0.0001) (Table 4).

Table 4.

Medication Dispensing Accuracy Comparison Between a Medically Staffed POD (2005 Exercise) and a Nonmedical Closed POD (2012 Exercise)

Group	Total (N)	Number (%) with Complete Dosing Information ^c	Number (%) Accurately Prescribed	Unadjusted Chi-Square P Value^d
All household members (both type and dose of medication)
2005 Exercise	2,112	1,879 (89.0%)	1,821 (96.9%)	<0.0001
2012 Exercise (both sessions)	745	696 (93.4%)	617 (88.6%)
Pediatric dosing accuracy^a
2005 Exercise	393	202 (51.4%)	181 (89.6%)	<0.0001
2012 Exercise (both sessions)	51	28 (54.9%)	8 (28.6%)
Accuracy for those with severe medication allergies^b
2005 Exercise	114	82 (71.9%)	79 (96.3%)	0.0623
2012 Exercise (both sessions)	109	109 (100.0%)	96 (88.1%)

In the 2005 exercise, pediatric dosing accuracy was assessed for those <11 years old. In the 2012 exercise, it was assessed for those <9 years old.

Severe allergies include rash, hives, impaired breathing, and swelling, and those with unknown allergy symptoms.

Complete dosing information to judge accuracy includes drug name, dose, and frequency of administration.

Chi-square value for comparison between 2005 public POD exercise data and 2012 closed POD exercise data; accuracy for severe allergies was not assessed by comparable methods between 2005 and 2012 exercises.

Discussion

This 2012 exercise provided the local health department with a real understanding of the ability of a nonmedical government agency to provide mass prophylaxis to its population with no onsite department support. This is a likely situation during a large-scale response when health department emergency responders may not be available for onsite assistance at all activated closed PODs. Overall, closed POD agency staff demonstrated that they could medicate their entire population in close to 24 hours. Throughput rates increased from session 1 to session 2. This was expected as training for session 2 was provided by health department staff and POD leadership roles were filled by department staff. In addition, closed POD staff became more familiar with their roles and gained a general understanding of closed POD operations in session 2, even though staff members were switched to fill different roles in the 2 sessions.

This exercise quantified error rates for dispensing the correct type of medication when contraindications existed and for pediatric dosing. This nonmedical closed POD had significantly higher error rates than the previously reported POD exercises staffed by health personnel, a finding consistent with previous research. We observed that just-in-time training provided by the health department decreased these error rates; however, they were still higher than error rates observed in PODs staffed by public health personnel. People with known severe allergies to classes of antibiotics are generally rare but do exist with respect to the medications used for a bioterrorist agent, and they may need a safer medication alternative. Our exercise demonstrated an 11.9% error rate for dispensing a safer medication alternative to a household member with a severe allergic reaction to the drug of choice, indicating that dozens of severe potentially life-threatening adverse events could occur in a POD of this size because of mistakes in processing. This error rate, however, was not found to be statistically significant when compared to the 2005 exercise in which health personnel processed the patients.

It is estimated that 1.3% of the population in the local jurisdiction is pregnant at any given time.^26,27 Based on this percentage, it is estimated that 301 pregnant women could be among the closed POD population. If the closed POD dispensing error rate for pregnant women is applied to the expected pregnant population of the government closed POD, an estimated 17 pregnant women would receive doxycycline (17=301*5.8%). Training or the inclusion of local health department staff did not significantly decrease error rates with respect to pregnant women receiving doxycycline, so other strategies must be explored to ensure that adverse events are minimized. One strategy that this health department is considering is rescreening pregnant women and switching them to a safer medication (eg, amoxicillin) upon provision of the additional 50-day supply to complete the 60-day course of prophylaxis when the biological agent is Bacillus anthracis.

It was expected that 12.6% of the population would be age 9 or under.²⁵ Applying this percentage to the closed POD population of 23,200, we predict that 2,923 children will require dosing adjustments. If the closed POD pediatric dispensing error rate is applied to the expected number of children who need medications, 2,087 will receive an incorrect dosing assignment. Small errors (within a half teaspoon) would still likely provide sufficient prophylaxis or result in minimal side effects due to overdoses; however, we saw that dosing errors might actually be substantial. This may require the health department to implement other strategies rather than relying on closed POD staff to provide the correct dose. To further decrease pediatric dispensing error rates, closed POD staff members provide a pediatric dosing information sheet to heads of households. The local health department has since updated the pediatric dosing guidelines to make them more user-friendly. Dose adjustments can be revisited at follow-up dispensing when the additional 50-day supply is provided, with a review of appropriate dosing for young children. Overall, the local health department intends to provide more intensive POD leadership training emphasizing vulnerable subsets of the population. We have discussed having screening supervisors periodically observe screening interactions from start to finish to ensure that algorithms are being followed correctly. Another strategy to ensure that patients receive the correct medication is to incorporate a quality assurance checkpoint prior to exiting the POD, where staff members review a randomly selected sample of patients' forms and medications, with emphasis on the aforementioned subsets of the population, to assess error rates and make corrections.

Limitations

We recognize that this exercise has a number of limitations. First, given the scheduling limitations of the closed POD agency, the local health department exercise team had a total of 3 hours and 15 minutes to conduct all activities, including registration, set-up, training, 2 drill sessions, and evaluation. In addition, the exercise team wanted to run 2 sessions so that all participants could experience the POD as both a staff member and a patient. Given these parameters, the brevity of the 2 exercise sessions was a necessity, although we acknowledge that a mean time frame of 23 minutes per session is not sufficient for producing optimal data regarding throughput and accuracy rates. Previous drills by this health department have demonstrated that personnel experience a learning curve throughout the duration of an exercise, in alignment with other findings.²² In other words, as staff become acclimated to their roles, error rates tend to decrease and processing rates tend to increase over time. Thus, throughput and accuracy during this exercise may have improved over time if the medical screening process had been prolonged.

A second consideration is that session 2 staff performance improvements may be overestimated and not completely attributable to health department training, as closed POD staff may have gained familiarity with the process during session 1. Another limitation regarding the analysis lies in the dosing error comparison between the 2012 closed POD exercise and the 2005 local health department public POD exercise. The 2005 exercise analyzed overall error rates and pediatric dosing errors using screening data, while the 2012 exercise used dispensing data, which may not be exactly comparable. In sum, data comparisons between the 2 sessions of the 2012 exercise and the 2005 exercise must be made with some care because of the differences in exercise design. Data from the 2005 exercise were included here merely to provide a baseline comparison for error and accuracy rates.

One final limitation is that given the scheduling limitations of the day, the local health department set up the entire closed POD in order to expedite that process. It is possible that if the closed POD had conducted set-up on their own, additional delays may have occurred and poorer throughputs may have resulted if the POD layout was executed incorrectly or POD stations were missing essential elements (eg, screening forms). We believe that, despite its limitations, this study offers valuable insight into closed POD operations and that the exercise outcomes highlight areas for further consideration and planning.

Implications

Potentially avoidable adverse medical events may be caused by improper screening or inaccurate dispensing of medicines. Asking nonmedical agencies to assume even limited medical responsibilities, including dispensing antibiotics, likely increases dispensing errors, thereby also increasing adverse events. This can affect subsets of the population, such as pregnant women and children, who are at higher risk for complications.²⁸ Furthermore, higher error rates among those who may be contraindicated may result in other adverse reactions of varying severity. Even nonsevere adverse reactions may result in discontinuation of antibiotic regimens, which may render prophylaxis ineffectual.¹² Underdosing of pediatric populations could mean that children are not sufficiently protected against developing a disease, while overdosing of pediatric populations may cause gastrointestinal problems and potentially life-threatening symptoms. Despite these potential adverse events, the severity of a public health emergency may warrant activating closed PODs and using nonmedical staff to prevent the widespread morbidity and mortality that would result from an anthrax attack or equally fatal exposure. Closed PODs are an operational strategy many health departments across the country intend to use, and decision makers should be aware of the consequences of activating this option.

Overall, we believe that these results support the feasibility of using closed PODs while highlighting the need for targeted training programs and follow up to reduce dispensing errors. Closed PODs are just one of many mass medication response strategies, and the decision to activate a closed POD will be based on the severity of the disease agent, the extent of exposures, the availability of medical countermeasures, and the timeframe to prevent illness. Though throughput rate is an important determinant of success in a mass prophylaxis scenario, accuracy should also be considered, as follow-up care for adverse events would further burden an already overwhelmed healthcare system.

Conclusions

This nonmedical closed POD demonstrated that it would be able to dispense emergency antibiotics to the heads of household of its entire internal population within the target timeframe. Nonmedical closed PODs likely have higher error rates of dose adjustments, although error rates for dispensing the correct medication and dose to children were decreased significantly following onsite just-in-time training from the local health department. However, this may not be a feasible response strategy if department staff are unavailable. Though closed PODs are a valuable strategy for reaching more people and conserving government resources, the associated reduction in accuracy should be considered when deciding to activate closed PODs during public health emergencies. When providing emergency prophylaxis, nonmedical closed PODs should be activated only if the consequences of delaying prophylaxis outweigh the expected adverse events.

Overall, the exercise provided the local health department with a first-time look at how closed PODs might perform with little or no training and demonstrated the capacity of a nonmedical closed POD to rapidly medicate its population. The error rates observed during this exercise further indicate the need for local health departments to conduct collaborative planning with the healthcare community to better define roles and capacity during an event requiring long-term mass prophylaxis. Depending on local capacity, ongoing or follow-up care for pregnant women, children, and patients who have other contraindications may be most safely and efficiently conducted by healthcare providers. In conclusion, there is still significant need to improve training materials for closed POD operations and to explore the capacity of the local health department to conduct patient follow-up, though the data support the expansion of the closed POD program at the local level.

References

Vietri

, Purcell

, Tobery

, et al. A short course of antibiotic treatment is effective in preventing death from experimental inhalational anthrax after discontinuing antibiotics. J Infect Dis, 2009; 199(3):336-341.

Hupert

, Bearman

, Mushlin

, Callahan

. Accuracy of screening for inhalational anthrax after a bioterrorist attack. Ann Intern Med, 2003; 139(5 Pt 1):337-345.

Day

. The autumn 2001 anthrax attack on the United States Postal Service: the consequences and response. Journal of Contingencies and Crisis Management, 2003; 11(3):110-117.

Martin

, Tierney

, Aranas

, et al. An overview of adverse events reported by participants in CDC's anthrax vaccine and antimicrobial availability program. Pharmacoepidemiol Drug Saf, 2005; 14(6):393-401.

SteelFisher

, Blendon

, Ross

, et al. Public response to an anthrax attack: reactions to mass prophylaxis in a scenario involving inhalation anthrax from an unidentified source. Biosecur Bioterror, 2011; 9(3):239-250.

Fowler

, Sanders

, Bravata

, et al. Cost-effectiveness of defending against bioterrorism: a comparison of vaccination and antibiotic prophylaxis against anthrax. Ann Intern Med, 2005; 142(8):601-610.

Davis

, Kammersall

, Altevogt

; Forum on Medical and Public Health Preparedness for Catastrophic Events, Institute of Medicine. Dispensing Medical Countermeasures for Public Health Emergencies: Workshop Summary. Washington, DC: National Academies Press; 2008.

Ramasamy

, Liu

, Tran

, et al. Principles of antidote pharmacology: an update on prophylaxis, post-exposure treatment recommendations and research initiatives for biological agents. Br J Pharmacol, 2010; 161(4):721-748.

Bayer. Ciprofloxacin [package insert].

10.

Pfizer. Doxycycline [package insert].

11.

GlaxoSmithKline. Amoxicillin [package insert].

12.

Meropol

, Chan

, Chen

, et al. Adverse events associated with prolonged antibiotic use. Pharmacoepidemiol Drug Saf, 2008; 17(5):523-532.

13.

Shehab

, Patel

, Srinivasan

, Budnitz

. Emergency department visits for antibiotic-associated adverse events. Clin Infect Dis, 2008; 47(6):735-743.

14.

Baccam

, Willauer

, Krometis

, Ma

, Sen

, Boechler

. Mass prophylaxis dispensing concerns: traffic and public access to PODs. Biosecur Bioterror, 2011; 9(2):139-151.

15.

Koh

, Elqura

, Judge

, et al. Implementing the Cities Readiness Initiative: lessons learned from Boston. Disaster Med Public Health Prep, 2008; 2(1):40-49.

16.

Walsh

, Van Groningen

, Craig

. Logistics modelling: improving resource management and public information strategies in Florida. J Bus Contin Emer Plan, 2011; 5(3):246-256.

17.

Aaby

, Herrmann

, Jordan

, Treadwell

, Wood

. Montgomery County's public health service uses operations research to plan emergency mass dispensing and vaccination clinics. Interfaces, 2006; 36(6).

18.

Lee

, Maheshwary

, Mason

, Glisson

. Large-scale dispensing for emergency response to bioterrorism and infectious-disease outbreak. Interfaces, 2006; 36(6):591-607.

19.

Hupert

, Xiong

, King

, et al. Uncertainty and operational considerations in mass prophylaxis workforce planning. Disaster Med Public Health Prep, 2009; 3 Suppl 2:S121-S131.

20.

Richter

, Khan

. Pilot model: judging alternate modes of dispensing prophylaxis in Los Angeles County. Interfaces, 2009; 39(3):228-240.

21.

Stergachis

, Wetmore

, Pennylegion

, et al. Evaluation of a mass dispensing exercise in a Cities Readiness Initiative setting. Am J Health Syst Pharm, 2007; 64(3):285-293.

22.

Spitzer

, Hupert

, Duckart

, Xiong

. Operational evaluation of high-throughput community-based mass prophylaxis using just-in-time training. Public Health Rep, 2007; 122(5):584-591.

23.

Agócs

, Fitzgerald

, Alles

, et al. Field testing a head-of-household method to dispense antibiotics. Biosecur Bioterror, 2007; 5(3):255-267.

24.

Ablah

, Scanlon

, Konda

, Tinius

, Gebbie

. A large-scale points-of-dispensing exercise for first responders and first receivers in Nassau County, New York. Biosecur Bioterror, 2010; 8(1):25-35.

25.

US Census Bureau. State & county quickfacts. http://quickfacts.census.gov/qfd/states/42/42101.html. Accessed April 29, 2014.

26.

Pennsylvania Vital Statistics 2004. Pennsylvania Department of Health. 2004.

27.

Centers for Disease Control, Division of Reproductive Health. Estimating the Number of Pregnant Women in a Geographic Area. http://www.cdc.gov/reproductivehealth/emergency/PDFs/EstNoofPregWom_508.pdf. Accessed April 29, 2014.

28.

White

, Henretig

, Dukes

. Medical management of vulnerable populations and co-morbid conditions of victims of bioterrorism. Emerg Med Clin North Am, 2002; 20(2):365-392, xi.