Design
A prospective observational study was performed to assess the temperature profile of thermolabile drugs once dispensed to ambulatory patients at a tertiary care hospital.
A continuous temperature measurement system was used to record thermal conditions from dispensation until the next outpatient visit.
Setting and participants
Son Espases University Hospital is a tertiary hospital located in Palma de Mallorca in the Balearic Islands, Spain. It is the reference hospital of a region with a population of more than one million inhabitants. The Pharmacy Department dispenses medications to over 7,000 patients every year. Patients were included in the study from July 2018 to January 2020.
Thermolabile medications
The following seven widely dispensed thermolabile medications were selected for the study in order to guarantee the number of patients needed and to avoid paediatric population and blood products: adalimumab, certolizumab pegol, darbepoetin alfa, erythropoietin, etanercept, trametinib, and peginterferon alfa 2-a.
Inclusion criteria
- Outpatients with a prescription of any of the selected thermolabile medications dispensed at the Pharmacy Department of Son Espases University Hospital.
- Signature of written informed consent.
- Older than 18 years.
Exclusion criteria
- Patients who do not give their consent.
- Patients participating in a clinical trial.
- Patients who presented some type of inability to understand the development of the study, such as a language barrier or intellectual disability.
- Patients who lived in residences and/or were admitted to nursing homes.
Intervention
Patients were recruited in the Pharmacy Department at the time of dispensing the thermolabile medication. When the patient came to the Pharmacy Department to pick up any of the medications included in the study, they were informed and offered the opportunity to participate. If the patient agreed, a questionnaire regarding the conditions that may affect the preservation of thermolabile drugs was completed.
Along with the medication, a data logger was added to the medication packaging. The data logger is a temperature recorder programmed to perform periodic measurements during the period between visits to the Hospital Pharmacy. If the patient was given more than one box, the container with the data logger was the last to be administered. The patient had to return the medication packaging and the data logger to the Pharmacy Department in the next visit, and the information was downloaded to analyse the data recorded. One medication was monitored for each patient. In case of detecting inadequate preservation, the importance of maintaining the medications in accordance with the recommendations of the Summary of Product Characteristics was emphasized.
Data logger characteristics
Temperature sensors TempTale®4USB, Sensitech INC, MA USA were used. Accuracy was ± 0.05ºC and calibration range was from -195ºC to 232ºC. Once started, temperature was recorded every 10 minutes.
Outcomes
The primary outcome was the proportion of TD improperly stored in the patients’ home setting. Temperature was considered inappropriate if one of the following circumstances were met: any temperature record less than or equal to 0ºC or over 25ºC; temperatures between 0-2ºC or 8ºC-25ºC for a continuous period over 30 minutes (three continuous measures).
Secondary outcomes included: proportion of TD with at least one temperature value less than or equal to 0ºC, proportion of TD with at least one temperature value over 25ºC, total time TD had been less than or equal to 0ºC, total time TD had been over 25ºC, total time TD had been within 8-25ºC, and total time TD had been over 8ºC.
As transport is the most sensitive period to temperature changes, we divided the data set into transport and home storage to analyse the thermal profile. The transport period was defined between the first temperature sample and the moment when the medication was placed in the refrigerator at home. The division between periods is based on a time series analysis algorithm defined by Bai J.11. The algorithm was implemented in R language by Marc Lavielle in 2017 and used in this study to find tendency changes in thermal profile12. This algorithm was applied using different considerations: first, considering the refrigerator works properly, therefore the home storage period will produce a smoother, less variable thermal profile than during the transport period. Second, related to the maximum transport period duration, since the patient is expected to store the medicine in the refrigerator within the first 24 hours from the time it was picked up. In Lavielle M. (2017), the mathematical model of the algorithm applied assumes that change point is the time point when the least-square method minimizes the residual sum of squares. Hence, the algorithm implemented splits the time series between transport and home storage periods where the residuals sum of squares of each period is minimum. The complete detail of this mathematical procedure can be found in Lavielle M12.
Additionally, the research team added 1 hour to the instant time obtained by the algorithm in order to include in the transport period the time needed by the medicine to reach the refrigerator temperature. Consequently, the transport period started when the patient received the medicine and the data logger was started, and ended when the algorithm detected the change point in thermal profile plus one hour more.
Medications were categorized sequentially into four exclusive groups:
Group 1: TD has at least one measure less than or equal to 0ºC.
Group 2: TD has at least one measure over 25ºC.
Group 3: TD has at least three continuous measures between 0º and 2ºC or between 8º and 25ºC.
Group 4: All other medications considered to be properly stored: measures between 2ºC and 8ºC with less than 3 continuous data out of this range but no data over 25ºC or less than or equal to 0ºC.
Sample size
Drugs dispensed to outpatients at the Pharmacy Department were analysed over 6 months, and 1,403 patients picked up TD, 625 of which were medications included in this study. Based on data in the literature, it was difficult to estimate the proportion of TD improperly stored, so we considered 50% to maximize the population.
A sample size of 107 randomly selected subjects was sufficient to estimate with 95% confidence and a precision of +/- 10 percent units, a proportion of TD improperly stored considered to be around 50%. A replacement rate of 25% was anticipated.
Data Analysis and Statistics
The recorded data were analyzed using R software version 3.6.3 on a Linux platform with the RSTUDIO IDE as user interface on real thermal data. The time series of temperature measurements obtained from each data logger were analyzed as statistically independent variables as each data logger was attached to different medicines and stored in different refrigerators and the total period of measurements were not the same in all cases, in some cases during winter and other cases during summer.
The data shown in the results section did not undergo any statistical treatment and must be considered directly related to thermal measurements.