Problem

Data from the US National Vital Statistics System show that the natality rate has remained stable since 1990. However, there has been a 50% increase in caesarean deliveries throughout the same period, corresponding to more than 30% of births.1 Consequently, caesarean delivery has become the most common inpatient surgery in the USA.2 Pregnancy occurs in both healthy and young populations. Therefore, it is reasonable to assume that caesarean delivery is the first inpatient experience for most patients.

With the current opioid crisis, healthcare providers are urged to personalise perioperative pain management, especially adjusting opioid prescriptions according to the patient’s needs. Although several mechanisms are involved in addiction, studies have demonstrated an association between perioperative opioid prescriptions and their persistent use, especially in opioid-naïve patients.3 In addition, growing scientific evidence shows that minorities are less likely to receive analgesics during hospitalisation, which is also true for caesarean deliveries.4 5

To better understand our institutional practices, inpatient opioid consumption after caesarean delivery was audited in the last quarter of 2017. The results showed important discrepancies in opioid prescriptions between different providers and an average opioid consumption that was twice the average reported by a previous study.6

Available knowledge

A previously published study concluded that patients undergoing caesarean delivery were three times more likely to fill an opioid prescription and 25% more likely to become persistent opioid users than those who underwent vaginal delivery.7 Another study revealed that filling an opioid prescription in the peripartum period was associated with increased persistent opioid use.8 A possible reason for the increased risk of persistent opioid use among these patients was investigated by Osmundson et al.9 In this study, 75% of the patients who received an opioid prescription after caesarean delivery had leftover tablets, which became uncontrolled sources of domestic opioids.10

Interestingly, the incidence of severe acute pain after caesarean delivery is lower than that after similar surgical procedures and may be related to physiological changes during pregnancy, such as increased levels of oxytocin.10 11 The development of chronic pain was also found to be very uncommon and was reported to be proportional to postprocedure pain intensity and not the delivery method.12

Given the risk associated with persistent opioid use in naïve patients and the usually low analgesic requirement after caesarean delivery vs similar surgeries, some institutions have implemented multimodal analgesic protocols that have successfully reduced overall opioid consumption.13 However, most previously published studies have compared differences in average and not according to trends over time, making studies susceptible to maturation bias and autocorrelation.14

Additionally, clinical trials have demonstrated that individualised opioid prescriptions reduce consumption without changing pain outcomes.15 However, two studies suggested that there is an inversely proportional relationship between opioid prescription and pain levels after caesarean delivery between different races.4 5

Rationale

Given the previously presented literature, it is reasonable that a personalised analgesia regimen can reduce both inpatient opioid consumption and racial disparity in opioid consumption. Therefore, our goal was to create a new opioid stewardship programme for caesarean delivery using minimal resources within our healthcare system. The goals of the opioid stewardship programme were to contextualise the opioid pandemic problem in the obstetric population for our providers, to reduce consumption using the new multimodal analgesia order set and to devise a clinical decision support system for pain management.

Specific aims

Our project aimed to decrease inpatient opioid consumption by 50% within 6 months of the opioid stewardship programme for caesarean delivery implementation. In addition, we analysed the possible impact of our programme in reducing the differences in opioid consumption between different races after caesarean delivery.

Context

The hospital where the quality improvement project was implemented is a 500-bed community hospital with an average of 4000 deliveries per year, of which 1000 were caesarean sections. Given the busy maternity workload, the hospital has several obstetric teams, further complicating uniform inpatient pain management practices and the potential risk of persistent opioid use.

Interventions

To create an opioid stewardship programme, the leadership of the anaesthesiology department, obstetrics department, nursing, pharmacy and medical informatics were involved. The opioid stewardship programme was initiated in March 2019 and was based on continued education lectures, new multimodal analgesia guidelines and a clinical decision support system. Lectures were performed monthly from March 2019 to August 2019 in grand rounds to increase awareness of risks related to opioid prescription in the obstetric population and to explain the new multimodal analgesia guidelines and the clinical decision support system.

The new multimodal analgesia order set consisted of a pre-established order set with scheduled acetaminophen (1000 mg every 6 hours) and non-steroidal anti-inflammatory drug (ibuprofen 600 mg or ketorolac 15 mg every 6 hours) administered alternatively every 3 hours while patients were awake. For breakthrough pain, an additional dose of a nonsteroidal anti-inflammatory drug (ketorolac 15 mg) and/or oral opioids (oxycodone 5 mg) was administered as needed. The clinical decision support system was embedded in the electronic medical record system as a preset order set, where non-opioid analgesics were prioritised in case of breakthrough pain.

Additionally, patients and family members were approached by previously trained staff who explained the programme’s rationale and how it worked; however, they were not intimately involved in the programme design or implementation. Patient feedback was important to improve how we addressed the opioid stewardship programme, especially in lay terms. In addition, patients were encouraged to disseminate their experience to motivate other expecting mothers during and beyond the programme.

Study of intervention

To evaluate the opioid stewardship programme, we started an inpatient opioid consumption analysis 12 months before programme implementation, and after that, a plan-do-study-act (PDSA) framework was employed. After initial planning and implementation (plan-do), inpatient opioid consumption was analysed as a trend over time using regression models and as an aggregate using monthly summarised data (study). If the expected reduction of opioid consumption was not achieved, the programme was reevaluated and adjusted accordingly (act). We used a cause-and-effect diagram to pinpoint the possible causes and then adjusted the PDSA cycles.

In addition, we analysed the results using an interrupted time series to compare opioid consumption before and after the opioid stewardship programme implementation and also with another hospital from the same healthcare network with a similar obstetric workload as a control.

Measures

Inpatient opioid consumption data were obtained from the pharmacy department after an audit of the electronic medical record system. Since the prescription and dispensation of opiates and benzodiazepines are connected to the electronic medical record system and the State of Ohio automated reporting system, data regarding opioid use have good validity owing to several audit processes involved, such as the witness requirement for medication waste and random inventory count. We excluded patients with a length of surgery longer than 3 hours, a length of stay of more than 7 days and total inpatient opioid consumption greater than 500 mg of oral morphine equivalents. For standardisation, all opioids were converted into oral morphine equivalents.16

Analysis

After validation and standardisation, inpatient opioid consumption data were analysed in two domains: consistency and differences in opioid consumption. Consistency was evaluated by the slope of the regression model, with values close to 0 showing a stable trend over time. Differences in opioid consumption were assessed using the ratio of geometric means since opioid consumption presents a lognormal distribution.

To evaluate the influence of healthcare evolution (maturation bias), we compared the regression before and after the opioid stewardship programme implementation and the control institution. Durbin Watson test was used to ensure that inpatient opioid consumption values over time were independent (autocorrelation).

Summary

Throughout the study, an 80% reduction in inpatient opioid consumption after caesarean delivery was achieved. The time series analysis demonstrated that the detected reduction in inpatient opioid consumption was associated with the opioid stewardship programme since the level achieved was significantly lower than the observed decreasing trend. This finding is reiterated by the inpatient opioid consumption observed in the control institution, where a reduction trend was observed, but on average, 31 mg oral morphine equivalents higher than the opioid stewardship programme hospital. Interestingly, African Americans received twice the amount of opioids compared with Caucasians throughout the study period, despite the reduction in inpatient opioid consumption for both races.

Interpretation

We believe that the results achieved by the opioid stewardship programme are secondary to the programme implementation given the chosen quasi-experimental method (segmented regression analysis).

Despite the simplicity of our opioid stewardship programme, a large retrospective study by Reed et al 16 demonstrated that the majority of patients did not receive multimodal analgesia after caesarean delivery in the USA.16 In addition, current evidence shows that different interventions can effectively reduce postpartum opioid use, but there is no optimal method yet.17 Therefore, we believe our programme applies in different contexts but requires continuous evaluation and adjustments over time.

One key lesson from this project is the importance of reaching out to all stakeholders. We believe that one of the primary reasons for the delay in achieving our goals was the peculiarities of the healthcare industry—large teams in different departments with distinct shifts and schedules. New programmes are more likely to be successful if all team members are aware of and motivated toward a common goal. In our project, the PDSA cycle was crucial for detecting a lack of communication with mid-level providers. The subsequent PDSA cycle confirmed our hypothesis, and we attained the initial goal of reducing inpatient opioid consumption.

Another aspect of the project dealt with changes in clinical practice. Despite the initial enthusiasm, some teams and patients were sceptical about the efficacy of our opioid stewardship programme. To overcome this challenge, periodic results presentations and positive reinforcement, such as public acknowledgement of team efforts, are needed. However, the way we presented the programme and its results was ineffective because some information was irrelevant to all groups. Perhaps we could have improved adherence if we asked local leaders involved in the opioid stewardship project to present their points of view to their respective groups.

Limitations

Our study had several limitations, such as the relatively short time frame analysed, which could be insufficient to observe the expected results, mainly when changes in clinical practice are expected. Analysing longer periods decreases the influence of maturation bias and historical events on the results.15 To address this problem, we used a controlled interrupted time-series method, taking advantage of the acquisition of a community hospital in 2017 by the Cleveland Clinic network. The data regarding inpatient opioid consumption after caesarean delivery in this new hospital were used as a control since the faculty and clinical protocols were independent, and both institutions were located in the same region.

Another limitation is the absence of patient perspectives on opioid stewardship programmes. In a patient-centred care model, our programme is reasonable only if it can reduce opioid consumption while providing physical comfort and emotional well-being to patients. Therefore, pain scores or anecdotal reports cannot capture a patient’s perspective on hospital care. More comprehensive assessments, such as the Hospital Consumer Assessment of Healthcare Providers and Systems, can be used to better understand patient perspectives on quality improvement projects.

We tried to find evidence regarding racial differences in opioid requirements after caesarean delivery since our study population had a fair proportion of African Americans (n=807; 33%) compared with Caucasians (n=1484; 60%). Interestingly, we observed that African Americans required more inpatient opioids than Caucasians during the entire study period, which is a conflicting result compared with previous studies. Although we cannot assume that this difference is based only on race since other confounders, such as comorbidities and pain scores, were not studied in our quality improvement project, we believe our results are interesting for future research.

Lastly, the COVID-19 pandemic changed the operations of many hospitals starting in March 2020. Given the magnitude and disruption in healthcare, data after February 2020 became less consistent due to health services reorganisation to deal with the pandemic. Despite all adversities imposed by the public health emergency, the opioid stewardship programme was continued and expanded to other facilities. Unfortunately, there are insufficient data to determine whether similar results were achieved.