CARES is a pragmatic parallel arm randomised controlled trial designed to compare the effectiveness of two analgesic prescribing strategies to treat acute pain after discharge from outpatient surgery at 1 week and up to 6 months after surgery. This study compares two flexible prescribing regimens that can each be tailored to individual patient preferences. Patients are assigned to one of two arms, either (1) the NSAID arm, which emphasises one type of NSAID non-opioid medication taken with acetaminophen, or (2) the opioid arm, which emphasises the use of one opioid analgesic taken with acetaminophen. CARES randomises adults undergoing one of three common low-risk surgical procedures, laparoscopic cholecystectomy, inguinal hernia repair and breast lumpectomy, to either the NSAID or opioid arm using a 1:1 allocation. The intervention structure and data collection protocol are the same for both arms; only the prescribing strategy assignment differs between them. The CARES study group is described in online supplemental appendix. Multiple trade-offs between pragmatic/effectiveness and explanatory/efficacy trial designs were considered, with a full PRECIS-2 assessment to characterise these choices across nine domains (figure 1, online supplemental exhibit 1).

For inclusion criteria, adults are eligible for participation if they have no significant analgesic medication use before surgery and undergo one of three common outpatient surgical procedures using the following definitions:

As a pragmatic trial, this study enrols a generalisable sample of surgical care patients who would be considered eligible for either NSAID or opioid analgesic therapy. We have therefore kept exclusion criteria to the minimum necessary to ensure both patient safety and internal validity. Adults who meet any of the following exclusion criteria that may interfere with providing informed consent or outcome assessment are ineligible: (1) schizophrenia, bipolar disorder or other psychosis; (2) anticipated other surgery within 6 months and (3) anticipated life expectancy of less than 6 months. We also exclude patients with absolute contraindications to either prescribing strategy.

Finally, because participants should be considered eligible in clinical practice for either NSAID or opioid analgesic regimen, patients with contraindications to acetaminophen, study-related NSAID drugs and study-related opioid drugs are excluded. In general, contraindications for specific medications include known allergy, liver disease or failure, estimated glomerular filtration rate (GFR) <60 mL/min, heart failure, peptic ulcer disease, anticoagulation, heart surgery, acute psychiatric instability (defined as current uncontrolled severe depression, severe post-traumatic stress disorder or suicidal ideation), substance use disorder not in remission or treatment and diversion of controlled substances.

Potentially eligible patients of participating surgeons are identified through local searches of preoperative visits in surgical clinical schedules, anaesthesia clinics and/or local electronic health record searches for eligible patients. Searches are updated at least every month during the enrolment period.

Potentially eligible adults are contacted via phone, patient portal or in clinic with a recruitment message describing the study. Potential participants may be contacted again within a week after receipt of this message to determine interest in participating and assess eligibility. If the patient is eligible and interested in participating, a study staff member guides the patient through the screening and informed consent process.

Potential participants are directed to complete a screening survey to determine eligibility for the study (eg, assess criteria like analgesic use before surgery). Potential participants may opt to either download the MyDataHelps study app for consent and screening or complete a parallel web version. Research staff provide technical support in downloading the app used for self-administered tools and patient-reported outcomes, and check for completion of surveys. Mobile devices are provided to participants who do not have one and use a web interface to assess any participants waiting for their device to arrive. After participants provide signed informed consent and authorisation, the baseline survey is conducted via the mobile app. A second method of enrolment is in-clinic contact of potential subjects by cross-referencing the potentially eligible list with the clinic and/or surgical schedule for each participating surgeon.

This study is performed at sites where we expect approximately 50–60% of the potentially eligible population for all three procedures to be women. Eligible women and those in under-represented racial and ethnic groups are encouraged to enrol. Initial enrolment targets are anticipated to be balanced across sites, with flexibility to modify targets as the study progresses.

Participants provide informed consent and complete the baseline assessment before surgery. On the day of surgery, patients are randomised to the NSAID or the opioid arm. Randomisation is stratified by type of surgery (laparoscopic cholecystectomy, inguinal hernia repair, breast lumpectomy) to assure balanced numbers of participants undergoing each procedure. Randomisation occurs in randomly varying block sizes of 2 and 4.

Participants are not masked to treatment arm assignment due to the complexity of the masking process and patient desire to know prescribing strategies. The supervising clinical investigators who implement the interventions also are aware of treatment assignment. To maintain allocation concealment, randomisation is conducted centrally using the electronic data capture form that captures the study arm assignment, using an algorithm prepared by the study statistician. Outcome assessors, who are research team members that may facilitate completion of surveys for participants in certain circumstances, are masked to treatment assignment. The masking of outcome assessors and the structured nature of outcome measures are expected to minimise potential biased ascertainment.

The analgesic prescribing strategies in this proposal were developed from evidence-based recommendations from analgesic prescribing guidelines and systematic reviews. Each medication has demonstrated efficacy from randomised controlled trials.2 Both prescribing strategies are operationalised by the surgical team, who provide the intervention in the form of analgesic medications prescribed to the patient, for both arms, on the day of surgery. Importantly, both the NSAID and opioid analgesic prescribing strategies are titrated to clinical response rather than a specific type, duration or dose of treatment. This pragmatic treat-to-target approach more closely mirrors real-world practice, which tailors evidence-based treatments to patient-specific outcomes.

The NSAID analgesic prescribing strategy includes prescriptions by the surgical team for one non-steroidal anti-inflammatory drug and acetaminophen. For the NSAID prescription, the surgical team may choose among one of the following three options:

The opioid analgesic prescribing strategy includes prescriptions by the surgical team for one opioid medication and acetaminophen. For the opioid prescription, the surgical team may choose to prescribe among one of the following three options, which are similar in opioid potency by morphine milligram equivalents:

For participants in both arms, the surgical team prescribes acetaminophen 1000 mg by mouth every 6 hours around the clock for the first 3 days after surgery then as needed thereafter (total #20 doses). Participants experiencing poor pain control will be encouraged to contact their surgical team. Participants may receive additional pain treatments at the direction of their surgical team, including additional doses of medications in line with their study arms. For example, additional doses of NSAIDs may occur in the NSAID arm (maximum doses of ibuprofen 3200 mg, naproxen 1500 mg, celecoxib 800 mg). Additional doses of opioids may also take place in the opioid arm. If needed, crossover to medications from the other study arm will be permitted.

Participants are informed of potential interactions and safety considerations, in line with standard of care, at the time of analgesic regimen prescribing. At study assessments, use of study medications and over-the-counter medications is evaluated. Participants receive closer monitoring than is available in usual practice. The study permits other aspects of perioperative and surgical care per the usual practices of surgical teams and does not specifically direct the use of local anaesthetics provided by the surgical team in the operating room.

Recommendations for the prevention of constipation are provided to participants in both study arms at the outset as per the usual practice at each study site. Important adverse events (eg, bleeding) are evaluated by a study physician and reported to the patient’s surgical provider. Potentially urgent adverse events (eg, chest pain) are referred for immediate evaluation in the local clinic or emergency department.

Study measures include both patient-reported measures that are collected in assessments at baseline and as outcomes in the time period after surgery. Assessments require approximately 25 min at baseline, 3 months and 6 months; and approximately 15 min at 1 week and 1 month. A small, graded incentive (in increments increasing from US$10 to US$15, totalling US$80 for the completion of all surveys in the study period) is provided for each outcome assessment. Incentives of this magnitude offset costs of participation and are in the standard range used in our previous studies. No incentives are provided for daily monitoring of pain and analgesic use, which is conducted for clinical intervention purposes.

Patient-reported outcome measures and timing of administration are displayed in table 1. Although participants are asked to complete a substantial battery of patient-reported measures in this proposed study, we perceive this outcome assessment protocol to be both appropriately comprehensive and reasonable in terms of respondent burden, and patient partners supported this approach to assessment. Outcome assessment protocols of similar length have been well tolerated in multiple previous symptom management trials involving patients with pain after surgery. Participants in these trials have also reported that they appreciated the sustained attention to their symptoms and perceived benefit from therapeutic disclosure.

National guidelines for the conduct of pain clinical trials recommend assessment of core outcome domains such as pain intensity and function, global improvement, symptoms and adverse events.3 Each of these pain outcome domains is assessed with validated patient-reported measures.

1. Secondary outcomes include measures for other pertinent domains. PROMIS Pain Interference scale (SF-6a) is a validated and reliable instrument that can be completed in <2 min to measure pain-related function.8–11 A seven-grade patient-reported global impression of change rating assesses patients’ views of overall improvement or worsening pain.12 Analgesic use includes assessment for the analgesic prescriptions written by the surgical team via chart review and patient-reported medication use (pills consumed (NSAID, opioid, acetaminophen), any other pain treatments). This permits an examination of adherence to the study regimen and potential crossover to the other study arm. A checklist of other postoperative pain treatments adapted from the Michigan Surgical Quality Collaborative will assess for non-prescribed treatments for pain after surgery such as ice, heat and cannabis (table 3).13 14

The Michigan Body Map has been validated to measure areas of chronic pain over the body. This will include questions on the symptom severity index to enable calculation of a score representative of widespread body pain.15 16 The Michigan Body Map will also be adapted to measure locations of acute pain over the body that may be outside the site of surgery. PROMIS Sleep Disturbance V.1.0 has been validated as four items measuring sleep problems.17 18 Quality of recovery is assessed via the internationally validated Quality of Recovery-15 score, which details five domains about how well a person recovers after surgery.19 20 PROMIS Preference score (PROMIS 29+2 Profile V.2.1) is a health-related quality-of-life measure that has demonstrated validity.21 It provides PROMIS scores for Cognition (Cognitive Function and Cognitive Function Abilities), Depression, Fatigue, Physical Function, Ability to Participate in Social Roles/Activities, as well as Pain Interference (4a) and Sleep Disturbance. PROMIS Anxiety includes 4 questions about anxiety. The Tobacco, Alcohol, Prescription medications and other Substance (TAPS) Tool consists of a four-item screening for tobacco use, alcohol use, prescription medication misuse and illicit substance use in the past year and brief assessment.22 23 One question from the National Survey on Drug Use and Health, a national survey with established validity and reliability, assesses for opioid misuse defined as use that is more than prescribed, for non-pain-related reasons, or in a way not prescribed by a doctor.24 25 New prolonged opioid use, defined as filling ≥1 opioid prescription post-discharge between 4 and 90 days and also between 91 and 180 days, will be assessed by patient report.26 Patients will also be asked to report on healthcare utilisation at 1 and 6 months after surgery, including unplanned postoperative clinical interactions related to pain (patient messages, phone calls, non-routine clinic visits related to pain), emergency room visits and hospitalisations.

At baseline, patients complete questions on demographics,27–33 a comorbidity checklist derived from the Charlson index,34 treatment expectations (table 4), depression (Patient Health Questionnaire [PHQ]-2),35 anxiety (Generalized Anxiety Disorder [GAD]-2),36 the Pain Catastrophizing Scale (shortform 6)37 and preference on the return of study results. The local study team reviews a patient’s chart to examine for characteristics of the perioperative period, procedure and Post Anesthesia Care Unit (PACU) course. The local study team reviews a participant’s chart to assess for clinically important adverse events at the end of the study period. This review also assesses for unplanned postoperative clinical interactions related to pain, including patient messages, phone calls and non-routine clinic visits related to pain. This also assesses for emergency room visits, hospitalisations and 30-day complications after discharge measured using the American College of Surgeons’ National Surgical Quality Improvement Program definitions.38

The central study team assesses new prolonged opioid use by obtaining prescription drug monitoring programme data, if available. Data necessary for linkage includes name (first, last), date of birth (month/day/year) and gender.

A mature and integrated digital informatics infrastructure is employed in CARES using CareEvolution MyDataHelps. Data gathered by patient report via the study app MyDataHelps participant portal are integrated into the trial data. Research participants consent using the app, complete patient-reported outcome measures and receive return of results. If patients do not have a mobile device, they may use a study participant portal to complete patient-reported outcomes. As a final step to ensure complete follow-up, research coordinators may contact patients to assist with completion of participant-provided information if necessary. Sites use MyDataHelps to document patient screening, approach, consenting and enrolment using electronic case report forms (eCRFs). MyDataHelps includes customisable, data-driven eCRFs to capture data gathered by research coordinators at each site. Standardised data forms capture data from the perioperative period, including immediately before, during and after surgery. MyDataHelps has been used to document clinical quality projects and prospective observational research, and has met strict medicolegal, audit trail, electronic signature and disaster recovery requirements across federal and state regulations.

In pain clinical studies, dropouts often occur due to intervention-related factors (eg, adverse effects, lack of treatment efficacy) and missing data can pose a substantial threat to internal validity. CARES has several design features meant to enhance participant retention and reduce missing data. First, the interventions are flexible treat-to-target analgesic prescribing strategies that allow tailoring to each type of procedure for patients. Second, both study arms are active interventions, which should reduce differential dropout that can occur among patients assigned to a control arm. Third, randomised participants are strongly encouraged to complete all outcome assessments, and individuals who do not respond to the study app assessment receive in-app reminders and phone call reminders from study staff. Fourth, to reinforce participants’ sense of commitment to the study, participants receive quarterly CARES newsletter, including tips on pain management and positive news related to recovery from surgery. Finally, if participants are at risk of dropping out due to assessment burden, they are given the option of completing a minimum core assessment comprising primary pain (ie, BPI) and adverse effect (eg, symptom checklist) measures.

The sample size for CARES was calculated based on testing the superiority of NSAID versus opiod analgesic regimens on effectiveness for acute pain. Because the surgical community has considered pain and postdischarge prescribing guidelines in the context of specific types of surgical procedures, the trial is powered to test the superiority for each one of the three types of low-risk surgeries included in CARES (ie, gallbladder removal, inguinal hernia repair, breast lumpectomy).39–43 To account for multiple comparisons, a two-sided type I error of 0.016 after applying a Bonferroni correction was used to control the family-wise error rate of 0.05 with three procedure types. Also, 90% power testing was used for superiority (two-sample t-test) at the day 7 assessment, but achieved power in the generalized estimating equations (GEE) models will be improved by the repeated outcome measures. The outcome measure for BPI pain intensity has a mean of ~3.4 (SD of 2.1) in prior samples of low-risk surgical patients, and past studies suggest a difference of 1.0 represents a minimally important difference.44–46

For the superiority analysis, calculations in Stata estimated that 244 patients (or 122 per arm with 1:1 allocation ratio) were required to detect a difference in means of ≥1. Based on prior work, we conservatively assume 15% for loss to follow-up. This requires enrolling 288 patients per type of surgery. To account for the three types of surgery, the study anticipates recruiting a total study population of 900 (300 patients per type of surgery).

The CARES study is also powered to examine heterogeneity of treatment effects and detect differences for subgroup analyses. After testing for effect modification, tests for examining outcomes among a priori selected subgroups will occur based on differences in acute pain, opioid prescribing and access to pain care.41 43 47 These subgroups include (1) male/female patients; (2) age<65/≥65 years; (3) black/white patients and (4) urban/rural patients. We calculated power for the effectiveness of subgroups using the total study population and Bonferroni adjustment for eight prespecified comparisons (two each for sex, age, race/ethnicity and urban/rural, with p values<0.00625). We anticipate the smallest subgroup size to be ≥15% of the total sample.

The study team formalised the data safety monitoring plan and formed an independent data safety monitoring board (DSMB) for CARES in consultation with the funder. The CARES DSMB provides independent review of the study to ensure the safety of participants and the validity and integrity of the data. Its members include relevant context experts (eg, surgeon, anaesthesiologist, biostatistician with clinical trials experience, patient partner, etc) and a DSMB Chair. The CARES DSMB charter was written and approved by concerned parties. The CARES DSMB meets two to three times per year.

Plans include for data by treatment group to be seen by the CARES DSMB after enrolment of the first 100 patients, and after 25%, 50% and 75% accrual of the sample; the CARES DSMB has sufficient data to assess data integrity and to compare rates of adverse events between treatment arms that may raise safety concerns. The CARES DSMB may consider termination at any point based on unexpected safety findings; however, early stopping criteria are not planned due to apparent benefit since neither approach is experimental, both approaches are widely used and because substantial evidence from an adequately powered trial is needed to affect clinical practice. The initial CARES DSMB meeting focused on review and approval of the study protocol and its informed consent template. Thereafter, the CARES DSMB reviews primary and secondary safety outcome data, data quality, enrolment data and projections.

We have engaged and will continue to engage with patient partners, who are patients or caregivers with lived experience with each of the three included types of procedures as well as being prescribed and using one or more of the study medications for pain relief. We have also engaged with other important stakeholders to develop CARES for several years using a variety of approaches. We conducted a patient-preference survey among 42 persons before surgery, which indicated that 71% would join the study and be randomised.48 Survey results also informed the study design, in that many respondents indicated reticence in participating in the study were their analgesic regimen to be masked. In partnership with the Michigan Institute for Clinical and Health Research, we convened a Community Engagement Studio to share a synopsis of the study protocol and learn the perspectives of 12 patients with lived experience. In the studio, patients offered feedback regarding the primary and secondary outcomes for the study, including identifying two outcomes of key importance (pain intensity, adverse events), as well as the recruitment strategy and factors regarding whether they would participate. Further, one of the CARES study co-investigators has experience as a caregiver and leader involved in relevant community organisations.

CARES also engages with patient and public members by convening a Stakeholder Advisory Board that includes five patients partners who include patients with lived experience and seven non-patient stakeholders. The CARES Stakeholder Advisory Board meets 3–4 times per year, where members offer perspectives and feedback that have informed the design and conduct of the study, and informs the dissemination of results. For example, patient partners and members of the Stakeholder Advisory Board reaffirmed and clarified the primary effectiveness and safety outcomes for the CARES trial. Patient partners identified worst pain after surgery, pain happening on the earliest postoperative days (ie, postoperative day 1 or 2) and side effects from pain medications as among the most important and relevant outcomes based on their experiences. In collaboration with our patient partners and other stakeholders, priority was given to the following questions faced by surgical patients requiring prescription medications to treat pain: (1) Will the analgesic regimen that I receive adequately relieve my acute pain as I recover from my surgery? (2) Will my analgesic regimen influence my risk of experiencing an adverse effect as I recover from my surgery? And (3) Will my analgesic regimen influence the quality of my recovery, my ability to function free of pain, my quality of life, my risk of experiencing chronic pain and my risk of either misusing opioid medications or using other substances? Patient partners and stakeholders also helped to select additional measures based on patient-centredness, validity, reliability, sensitivity and brevity, with focus on reducing patient burden.

Collaboration with the Stakeholder Advisory Board, including patient partners, includes framing results from the CARES trial in ways that are understandable and impactful to the millions of patients who undergo low-risk outpatient surgeries every year. Co-production of three types of content and content summaries for the CARES trial is planned, which include lay language text summaries of overall study findings, contextual text summaries that include insights relevant to different study subgroups and digital multimedia summaries that use video, audio and other media.

Not applicable.