I mentioned in an earlier post that I would devote an exhaustive post to our Adherence project, so here it is! Forgive me if I have missed out on anything, so please see Kristen’s earlier posts as well.
Importance of adherence: Now adherence is defined as the extent to which a patient follows their prescribed regimen. Barriers to maintaining good adherence include difficult ARV side effects, no tolerance to the medications, and unavailability of money or transport to maintain the supply of medications. Inadequate food or water supply, no social support or means of refrigeration for certain medicines and complex timing schedules for administering ARV’s also act as a hindrance to maintaining a good adherence. Poor adherence can lead to progression of the illness, the development of resistance to the medications, increased healthcare costs, and even death. As such, it is important to check a patient’s adherence to their medication at each clinic visit to emphasize to the patient the importance of good adherence and to let the physician know if there are any medical concerns. For example, for a patient with poor adherence, a physician may recommend a viral load and CD4 test to assess whether or not the patient needs a change in drug regimen or whether the patient should be moved to the next line of ARV treatment (of which there are ONLY three lines).
Current methodology: In Botswana, adherence is measured in each examining room by a nurse. For pill medications, they make a manual count of the actual returned and compare this to the expected return to calculate a percent adherence. The nurse sits alongside the doctor in the examining room and reads out the calculated adherence for each medication to the doctor to be recorded on the Medical chart for the patient. For liquid medications, the nurse rinses out a tall graduated cylinder, labeled for the specific medication, and then pours the remaining liquid medication to determine the milliliters of liquid remaining. Finally, the nurse pours the medication back into the bottle (to return to the patient) and makes a note of the remaining volume for the pharmacy to dispense more. Finally, the cylinder is rinsed again. Since they use large graduated cylinder, there is a greater chance of losing liquids due to their highly viscous nature (i.e. taller cylinder means more surface area for the medication to get stuck to and then more difficult to remove). Moreover, using these graduated cylinders also increases the risk of spilling the liquid medications because it involves pouring medicine from bottle to cylinder and back to the bottle. It is also time-consuming as it involves extra steps of rinsing the cylinders and taking off the bottles caps and popping off the nozzles.
Now in Swaziland, there is an individual adherence room where all patient medications pass through first, before seeing the doctor. Basically, there is one nurse who does the pill counting and assesses the volume of liquid medication remaining. The number of actual pills returned and actual volume of liquid medicine remaining is then recorded on an Adherence Record Sheet and these numbers are also entered by another nurse onto the Electronic Medical Record system (EMR). This Adherence Record Sheet is attached to the patients file and then the patient returns to the waiting room to be called on to see the physician. Within EMR though, adherence is calculated using the following inputs: medication name, date of last clinic visit (entered using a drop-down calendar), doses per day, expected return volume (pill count), actual returned volume (or pill count). Using these inputs, EMR calculates the percent adherence for each medication. After discussing any issues of poor adherence with the patient’s family, the nurse can make notes within EMR that will be visible to the physician as they make notes during the examination. The Clinic here uses a series of smaller medicine measuring cups to determine the volume of liquid medication remaining and as such they lose a smaller volume of medication when measuring adherence each time. Spillage and time are still a problem though. Relying on EMR to calculate adherence is also a downside because EMR fails frequently and the internet connection downstairs is not very stable.
The Adherence Monitoring System: Interns from last year the Lesotho and Botswana clinic both mentioned that calculating adherence, though necessary, was quite tedious and time-consuming. Thus, our aim was to create a device for the Baylor International Pediatric AIDS Initiative (BIPAI) Clinics that measures adherence to anti-retroviral regimens in a time-efficient manner.
The system consists of two components: 1 Acculab VIC-303 scale and 1 Microsoft Excel computer program. The scale can measure up to 300 grams with 0.001g readability. After calibrating the scale, the average weight of each pill medication, of 1cc of liquid medication, and of the empty liquid medication bottles, must be catalogued into an Excel spreadsheet before using AMS.
The inputs required for the Adherence % Excel program include the present date (automatically updated), date of last clinic visit, medication name, bottle type, initial dosage, doses per day, and the remaining mass reading from the scale. Outputs include the number of days between visits, the recommended dose, the pill count (or volume) remaining, actual dose taken, percent adherence, and the number of pills (or mLs) missing or extra. We think this program will be especially useful in the event the electronic record system fails. The other Excel program takes the input of medication name, bottle type, and remaining mass to output the number of pills (or volume of liquid medication) remaining. This program can be used in conjunction with the electronic system to calculate adherence percentage.
To measure adherence for pill medications using AMS, the pills should be transferred into a standard tared container and the remaining mass can be entered into the Excel program. Most patients come in with less than 6 pills so in these instances it is faster for the nurse to just open the bottle and count the number of pills remaining. We decided against cataloguing each pill bottle because this means extra foil, cotton balls, or any other extraneous items would have to first be removed, before placing the bottle on the scale. If these items are not removed, then this can add more variability to the already present variance in empty bottle weights. For example, we found that for one medication, the variance in bottle weights was off by as much as 4 grams, which can throw the remaining pill count off by as much as [4grams/0.15grams (or the weight of the smallest pill)]. For this reason, we decided to just use a standard tared container to weight the mass of remaining pills alone.
To measure adherence for liquid medications using AMS, the bottle cap along with any nozzles should first be removed before placing the bottle containing the liquid medication onto the scale. This mass reading can then be entered into the Excel program to obtain the volume of liquid remaining and percent adherence. For liquid bottles, we wanted to avoid using graduated cylinders and taking out the liquid medication for any reason. Bottle weight variance of empty liquid medication bottles was still a concern but we decided the best way to overcome this was to weigh as many empty bottles as we can, so that the average mass entered into the Excel program is a good representation of the mass distribution.
Swaziland Results and Stats: As Kristen explained earlier, “the centralized location of measuring adherence in Swaziland makes it much easier to use our device because we only need one device (vs. one for every examining room in the Baylor clinic) and there is an available computer for us to put the program on (vs. in Botswana the doctor is on the only available computer in the room).”
Once we reach Swaziland, we spent one morning to observe how things ran in the Adherence Room. Soon afterwards, we decided to go ahead and show the nurses how AMS worked. They were very eager and excited about the prospect of not having to move liquids around anymore. Anyhow, for the few liquid medications we tried, AMS seemed to be off by as much as 3 - 20ml. This worried the nurses because for some medications, such as Kaletra, a reading off by 3 milliliters could cause the calculated adherence to be off by 3 doses. The nurses also seemed somewhat weary about using the scale and were quite concerned about what to do in the event of a malfunction. The fact that manufacturers frequently change bottles for liquid medication was also a concern.
We came in the next day and decided to basically re-catalogue all the medications used. We noticed that many of the bottles for both pill and liquid medications were different than those medicines used in Botswana, probably because they came from a different manufacturer. As we were cataloguing, we observed that the weight of 1cc of liquid medication or of the average pill was definitely different than that weight recorded for Botswana; and this could explain the reason why AMS seemed to be 3mls off on the previous day. After some exhaustive cataloguing and more testing, we were very relieved to find out that AMS was returning pill counts to the precise pill (i.e. AMS gives 5.04 when there are 5 pills left OR AMS gives 3.992 when there are 4 pills left). For medications where there were half pills, AMS returned 46.45 pills remaining when there were actually 46.5 pills; thus, AMS is also sensitive to the presence of half pills. A paired t-test results showed there was no significant difference for the remaining pill counts obtained through the manual method or AMS.
Now for liquid medications, AMS also proved promising again as we only observed a deviation between 0.1 to 2.0ml in calculating the volume of liquid medication remaining. A paired t-test results showed there was no significant difference in calculating the volume of liquid medication remaining (p > 0.05) however there WAS a significant difference in the time required to determine the actual volume remaining (p < 0.05). The average time for the manual methods was ~52 seconds while with AMS, this average was much faster at only ~24 seconds. Although AMS was quite accurate in providing the volume of liquid remaining, we recommended that for volumes below 25 milliliters or for medications with doses as low as 1milliliter, it was best to use the manual method since a gap of 2mls with AMS can lead to an adherence percentage off by as much as 2 doses. The effect of 2 doses missing or extra on the calculated adherence percentage depends on a number of factors including the number of days between visits and the initial dosage given. As Kristen mentioned, for the few examples in which we did adjust the volume remaining by 2mls, the adherence percentage only changed by 1%.
Moreover, the nurses in Swaziland were pleased with AMS’s performance last Friday so they are testing it out, alongside their manual method during this work week. Their feedback is quite important to us and we will let you know what they say!