Study Design
This randomized controlled trial was conducted in two serial phases, in which female patients between the ages of 18 and 50 years who were undergoing a cesarean section at Temple University Hospital were recruited between May 2021 and August 2022. During each phase, a block random assignment with a block size of 10 was used [19], assigning participants into one of two study groups—the CPMRx mobile app or the control group—and ensuring that 5 participants in a block of 10 will be assigned to each study group. The allocation sequence generation and subsequent enrollment of participants were completed by different members of the study team. Phase I was conducted from May 2021 to March 2022 and included the first 50 recruited participants. For Phase I, the control group was provided electronic monitoring blister packs for their prescribed pain medications (BP Only) and the treatment group was provided the CPMRx mobile application in conjunction with the use of the blister packs (BP + App). Phase II was conducted from March 2022 to August 2022 and included the next set of 50 participants recruited. In Phase II, the use of electronic monitoring blister packs was removed in order to decrease the patients’ perception of being monitored. The control group in Phase II had a hard count of their prescription pain medications conducted by a member of the research team (HC Only) and the treatment group was provided the CPMRx mobile application in addition to the hard count (HC + App).
The trial was approved by the institutional review board of Lewis Katz School of Medicine at Temple University (Philadelphia, PA) prior to recruitment of participants. The study was registered at ClinicalTrials.gov (NCT05461196, 18/07/2022). All participants provided voluntary written informed consent. All methods were carried out in accordance with relevant guidelines and regulations. The CONSORT checklist was used when writing our report [20].
Participants
There were multiple inclusion and exclusion criteria that were established for recruitment of our participants in the study. The inclusion criteria included: female adult between the ages 18-50 years, patient was undergoing a cesarean section procedure (both scheduled and unscheduled) at Temple University Hospital, patient was able and willing to provided informed consent, patient had insurance that would cover the cost of prescription drugs or was able to pay out-of-pocket, and the patient owned a smartphone (>80% of the target population owns a smartphone). Exclusion criteria included patients with contraindications to opioids, NSAIDS or acetaminophen (including but not limited to allergy, intolerance, history of prior misuse, inability to tolerate pills), a diagnosis of acute or chronic pain disorder, non-English speaking, unable to provide consent and if the patient was currently incarcerated.
Measures
Outcome Measures. The primary outcome for this study was the total number of prescribed opioids used and the secondary outcome was opioid misuse during the postoperative period. Opioid misuse was defined as any situation in which opioid use was outside of prescribed parameters (e.g., in greater amounts or more often than prescribed), and included unintentional misuse (such as misunderstanding of instructions) or possible aberrant behaviors (such as recreational use or diversion). Total number of prescribed acetaminophen and ibuprofen used were also obtained.
Independent Variables. Reviews of electronic medical records were conducted to obtain potential covariates and control variables, including sociodemographic characteristics, mental and physical health history, and clinical characteristics related to the cesarean section.
Sociodemographic characteristics included age (in years), race (White or European American, Black or African American, Asian American, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, Two or more races, or Unknown), ethnicity (Not Hispanic or Latina, Hispanic or Latina, Unknown), and type of insurance used (private insurance, Medicaid, uninsured).
Health history included body mass index (BMI), tobacco use, and diagnosis (ICD-9 or ICD-10 codes) of anxiety, depression, sleep disorder, alcohol abuse, substance use disorder (other than OUD), chronic back pain, and/or migraines. Current use of antidepressants, antipsychotics, and benzodiazepines were also documented. Postpartum depression was measured using the Edinburgh Postnatal Depression Scale [21], a 10-item self-report scale administered to all postpartum patients; the EPDS has a maximum score of 30 and higher scores indicate higher levels of postpartum depression; scores above 10 may indicate possible depression.
Clinical characteristics included whether the cesarean section was scheduled or unscheduled, whether the cesarean section was primary or repeat, whether a hysterectomy was performed at time of the cesarean section, the type of skin incision (Pfannensteil or midline vertical), and whether the patient labored prior to the cesarean section.
Intervention and Procedures
The participants in the CPMRx mobile app group (experimental group) downloaded the CPMRx mobile app prior to discharge and used it throughout the 10-day post-operative period to score pain levels and engage in gamification to delay opioid pain medication use, in addition to bringing all pain medications to their postoperative appointment for either a scan of the electronic monitoring blister packs or a hard count of the pills. The participants in the electronic monitoring blister pack or hard count-only group (control group) brought all prescribed pain medications to their postoperative appointment for either a scan of the electronic monitoring blister packs or a hard count of opioid pain medication usage, as well as acetaminophen and NSAID use.
At the time of discharge, the surgeon wrote a prescription for pain medications according to usual practice. The current standard of care postoperatively is that patients are instructed to take scheduled Motrin (dispense 30 tablets) and Tylenol (dispense 30 tablets, 650mg) with opioids (15 tablets Roxicodone, 5mg) for breakthrough pain control. The patient was directed to take this medication post-op as an outpatient. The medications’ (ibuprofen, oxycodone and acetaminophen) anticipated use is approximately 1 week postoperatively. Upon discharge, the patient self-managed their pain using the pain medications prescribed to them by the surgeon. During Phase I, all pain medications were dispensed in an electronic monitoring blister pack that recorded the exact time of use for each pill removed.
Power Analysis
Because of the pilot nature of the existing literature, sufficient information to calculate estimated effect sizes did not exist; consequently, we used an expected medium effect size for all aims. Sample size calculations revealed a required sample size of 80 participants for an analysis of variance (ANOVA) where the effect size f = .25 (medium), α err prob = 0.05, power (1-β err prob) = 0.80, and 2 groups (non-centrality parameter λ = 8.4, Critical F = 4.2, Denominator df = 98, Total sample size = 80, actual power = 0.801) 45. All other aims would require a smaller sample size, so we targeted enrollment of 100 participants, considering a 20% attrition rate.
Statistical Analysis
All analyses were run using SAS software, Version 9.4 (SAS Institute Inc., Cary, NC). To mitigate any potential perceived conflicts of interest, all analyses and raw data were independently verified by biostatisticians at Temple University outside of the study team. Descriptive statistics and covariates (see Table 1) were computed for the full sample and separately for each group. Bivariate group differences were assessed using linear regression (for continuous outcomes and covariates) or Wald chi-square tests for two-way contingency tables (for categorical outcomes and covariates). Linear and logistic regression analysis models tested the CPMRx treatment effect on each outcome in Table 1 controlling for the covariates in Table 1 that were significantly or near significantly different by CPMRx and control groups. Relative risk estimation by Poisson regression with robust error variance was used to determine if there were significant differences between participants who completed the study and those who were lost to follow-up. Logistic regression analyses were run to predict opioid misuse by group (control [electronic monitoring blister pack only] versus CPMRx [electronic monitoring blister pack plus mobile application]) and odds ratios (ORs) were calculated. Relative risk (RR) was also estimated by log-binomial regression.