Study design

This population-based, retrospective cohort study explores gender-related differences in persistent opioid use in previously opioid naïve patients presenting with renal colic. The present study expands on a parent study of linked administrative databases describing all opioid-naïve patients diagnosed with renal colic between July 1, 2013 and September 30, 2017 in Ontario [16]. As of 2012, all opioid prescriptions that are dispensed in Ontario are captured in the Narcotic Monitoring System (NMS). Administrative databases utilized were all linked through Institute of Clinical Evaluative Sciences (ICES) as previously described [16]. The accuracy of these databases in terms of quality and coding has been previously discussed [17]. Codes utilized in this study have been summarized in S1 Table and the team accessed the dataset between August 2019 and March 2020.

Exclusion criteria were patients that had been dispensed an opioid in the year prior to their renal colic presentation, those who had a renal colic episode in the year prior, and any patient who had received palliative care services either the year before or the year after their index presentation. This study was approved by the Queen’s University Health Sciences and Affiliated Hospitals Research Ethics Board. All data were fully anonymized before accessing them and requirement for informed consent was waived.

Outcomes

Persistent opioid use after renal colic by documented gender was the primary outcome. Persistent opioid use was defined by one or more opioid prescription(s) dispensed between 0 and 90 days following the index renal colic visit that subsequently resulted in one or more prescription(s) for opioids that was dispensed between 91 and 180 days following the index visit. These definitions are similar to that of other comparable studies [10, 11] although we also determined dispensed opioids 6–12 months and 1–2 years after index diagnosis. Secondary outcomes included any subsequent diagnosis of opioid overdose, opioid addiction, hyperalgesia, and opioid-related death [17, 18].

Covariates

Covariates were determined a priori and conceptualized as confounding variables such as age, income quintiles, geographic location, patient comorbidities (Charlson index) [19], and enrollment with a primary care practitioner (PCP). The majority of these explanatory factors were conceived, documented on a Dataset Creation Plan, for the parent study [16] including clinical care-related covariates such as number of emergency department (ED) visits, gender of prescriber, number of PCP visits, and total oral morphine equivalents (OMEs) [20] dispensed during the index presentation. Given past literature focusing on sex and gender on pain perception, experience and management [3, 14, 15] we also included new covariates: mental health utilization, substance abuse and disability program registration (ODSP). The most important covariates not available in this dataset are the kidney stone details (size and location).

Statistics

Descriptive statistics (ratios) were used for demographic and baseline characteristics and to compare proportions between groups the Chi-square test was used. To identify the factors associated with the primary outcome, both univariate and multivariable logistic regression models were used. All patient-related and care-related variables were included in the original models and then subsequent iterative models were performed with groupings of covariates as described. Given the previous literature around gender, mental health diagnoses and pain outcomes we performed a test of interaction terms of gender and any mental health services utilization with the primary outcome of persistent opioid use. Missing data was included in the analyses as unknown indicator variable. A two-sided p-value of <0.05 was considered statistically significant. As per institutional policy, cells with <6 patients were not reported due to privacy concerns. Data were analysed using SAS Stat 14.3.

Interpretation

In this population-based study, we demonstrate gender-related differences in persistent opioid use beyond 3 months after an intended short-term course required for renal colic, independent of multiple key confounders or explanatory variables. Of the 2,236 (9.6%) previously opioid naïve females that continued to fill opioid prescriptions after an acute pain event, more than a third had evidence of persistent opioid use 1–2 years later. Females were more likely to have documented utilization of mental health services compared to males in this cohort and given the strength of its association with persistent opioid use, it could be hypothesized as an explanatory factor although gender continued to be significant even after adjusting for this covariate. Key treatment-related factors associated with persistent opioid use including need for surgical intervention and the amount of opioids dispensed appeared to be similarly associated in both males and females.

Female patients in this cohort, representing all those diagnosed with renal colic during the study period, differed in several parameters compared to their male counterparts, some of which have been previously associated with continued opioid utilization. Female patients tended to be somewhat younger with some evidence of lower SES, although other statistically significant variances in baseline characteristics such as rurality, comorbidity and previous stone surgery were likely not clinically different. Nonetheless, previous studies exploring associations of persistent opioid use and patient-related factors would suggest many of these variables such as age, SES and comorbidity may confound these findings [10, 11]. Female patients were less likely to be prescribed an opioid during their renal colic presentation (62.1% vs. 67.9% for men) and the total OME prescribed to women was significantly less (23.8% females prescribed >200 OME vs. 28.7% for males) which is consistent with previous studies. This finding is interesting given the strength of the association of dose and duration of original opioid prescriptions and persistent use in otherwise unselected patients with acute or chronic non-cancer pain [21]. As well, females appeared to have slightly earlier resolution of their renal colic event and more likely to have earlier stone related surgery. Despite these findings that might predict less ongoing opioid utilization, females still demonstrated more opioid use 3–6 months after the acute pain event in both unadjusted and adjusted analysis.

There were other apparent gender-related differences in the management of patients with renal colic which may have contributed to the primary outcome including more surgical interventions (23.2% for males vs. 26% for females) and more health provider visits during the initial renal colic event which may have been associated with repeated access to opioid prescriptions. We have previously reported several kidney stone surgery specific risk factors for persistent opioid use [16] including an increased number of interventions as well as surgical type. These factors appeared to be similarly associated with the primary outcome in both males and females in this present study (S3 Table).

Multiple studies have investigated differences in opioid prescribing and opioid use between genders with consistent findings that characteristics such as mental health diagnoses and post-traumatic stress are strongly associated with opioid utilization in the chronic pain setting [2, 3, 14]. Similarly, a limited number of recent studies exploring persistent opioid use following acute pain events in previously opioid-naïve patients have found similar results in terms of gender-related risk factors [21, 22]. Involvement in a disability program, such as the ODSP in this present study, has been also been shown to be associated with long term opioid use [23], but few have included such programs in their analyses of persistent opioid use after short-term acute opioid prescriptions. In this current study, mental health services utilization appeared to be strongly associated with continued opioid use in both males and females in adjusted analysis. However, the incidence of such diagnoses was significantly higher in females than their male counterparts.

Given the millions of low-complexity procedures that are performed every day, in addition to the various acutely painful conditions that present to primary care and urgent care clinics, these significant rates of persistent opioid use are concerning. The strong association of previous mental health services utilization and continued opioid use is consistent for both males and females, however the increased incidence of a mental health diagnosis in females in this large population-based cohort emphasizes their vulnerability. The specific etiological basis underlying any gender differences in pain experiences are complex and multi-factorial, and emerging evidence suggests endogenous opioid functioning may play a causal role in these disparities, including sex hormones as factors influencing pain sensitivity [1]. Psychosocial processes, stereotypical gender roles manifesting in differences in pain expression and sex differences in responses to analgesic medications have been implicated [1, 5]. Despite adjusting for multiple key factors associated with persistent opioid use in this present study, the fact that female gender remained an independent variable associated with persistent opioid use underscores the need to consider gender when identifying vulnerable populations with regard to opioid prescribing.

This study is limited by the retrospective nature as well as the accuracy of the data used, including diagnostic codes, procedure codes and opioid prescription capture, leading to potential misclassification bias. Furthermore, it is not possible to determine potential additional diagnoses for which ongoing opioid management would be considered, or the severity of the diagnoses that were coded. In addition, the NMS only tracks opioids dispensed, not necessarily those actually utilized by patients, or opioids utilized by a patient from a non-physician source (e.g. friend/relative’s prescription, street). Importantly, information around the size and location of urolithiasis was not captured in the current study. Finally, there remains the potential threat of residual confounding. Strengths of the study however include the large population-based cohort size and the ability to determine opioid utilization prior to the urolithiasis diagnosis as the NMS captures opioid prescriptions for patients of all ages in this mandatory, province-wide reporting system.

In conclusion, the majority of patients diagnosed with renal colic in this real-world study were prescribed opioids, despite increasing awareness of risks and documented efficacy of non-opioid medications. Females were at slightly higher risk to fill prescriptions beyond 3 months after the acute pain event and this remained significant even after adjustment for key confounders. The data support and highlight the urgent need for opioid reduction strategies, including opioid-reduced prescriptions and multimodal non-opioid analgesia strategies with a focus on provider education to identify vulnerable populations.