Preoperative Medication Management, Compliance and Adverse Events in Adult Patients Undergoing Elective Surgery: A Historical Chart Review ()
1. Introduction
The anesthetic preoperative evaluation is an essential step in a patient’s journey to the operating room. One of the main goals of this process is to reduce patient risk and morbidity [1]. This process includes a medication reconciliation, where the anesthesiologist informs the patient about which their current medications should be continued or discontinued prior to surgery, and which should be taken on the day of surgery.
The American Society of Anesthesiologists (ASA) has published a Practice Advisory for Preanesthesia Evaluation; however, this advisory does not address medication reconciliation in the process. The literature includes a plethora of studies and guidelines suggesting preoperative alterations to patients’ medications. The recommendations for preoperative medication instructions are constantly evolving and must be applied by balancing the risks and benefits for the individual patient. Many of the medications examined in the literature are cardiovascular (e.g., antihypertensives and anticoagulants), although narcotics, steroids, antipsychotics, and hyperglycemic medications, among others, are also considered by anesthesiologists during the preoperative evaluation.
Medication compliance, particularly for chronic medications, has been well-studied in the literature. Despite their consistency, the findings are often vague, for example, an oft-quoted statistic concerning patients’ medication compliance states that the adherence rate is around 50%, ranging from 0% to 100% [2].
There is a gap in the literature, however, regarding medication compliance in the perioperative period. Most studies tend to focus on physician-led errors. Two studies demonstrated that the addition of a pharmacy presence in the preoperative clinics may mitigate the providers’ errors in medication reconciliation [3] [4]. Only a few studies examined patient medication compliance in the preoperative period, two of which report non-compliance rates of 22% and 65% [5] [6]. Notaras et al. found that common reasons for patients inappropriately omitting medications in the preoperative period included not being administered the medication by ward staff, not being medicated for the indicated condition, not understanding instructions, or not remembering [6]. Two studies demonstrated that the use of standardized instruction sheets may improve patient medication compliance in the preoperative period [7] [8]. Interventions to improve medication compliance may be effective, but are often complex and labour-intensive [9].
The present study sought to determine the proportion of adult patients undergoing non-emergent surgery who are non-compliant with preoperative medication instructions and identify predictors of non-compliance. The secondary objectives were to: 1) describe medication instructions by drug type; 2) explore the impact of non-compliance on blood pressure among patients receiving various antihypertensive medication; and 3) investigate the relationship between medication non-compliance and intraoperative/perioperative: a) hemodynamic adverse events, b) glycemic adverse events, c) thrombotic adverse events, d) respiratory adverse events, and e) patient disposition.
2. Methods
The present historical chart review of 393 randomly selected anesthetic records was conducted at Hamilton Health Sciences (HHS) after obtaining approval from the Hamilton Integrated Research Ethics Board (HiREB 13-442-C), which is the research ethics board that oversees research projects involving McMaster University, St. Joseph’s Healthcare Hamilton, and Hamilton Health Sciences. The HiREB is responsible, on behalf of the institutions, for ensuring that all research involving human subjects under the auspices of its institutions meet current ethical standards. HHS is a network of seven hospitals and a cancer centre in Hamilton, Ontario, Canada.
Data Collection
Adult patients who underwent elective surgery at HHS between May 1, 2012, and April 30, 2013 were included. These patients undergoing elective surgery attended a mandatory HHS preoperative clinic. This centralized preoperative clinic serves three hospitals: McMaster University Medical Centre, Hamilton General Hospital, and the Juravinski Hospital. The HHS preoperative clinic processes over 100 patients per day and was staffed until 4 p.m. by at least four anesthesiologists (one pediatric specialist, three adult specialists, and occasionally a resident physician), as well as a nurse and pharmacist. In these clinics, the pharmacist met with the patients and prepared a list of their medications on the paper anesthesia record. In some cases, the medication list was collected by a nurse or the anesthesiologist when a pharmacist was not available. Subsequently the patients were seen by an anesthesiologist who did a consult that included medication management. The anesthesiologist was charged with telling the patient what to do with all their medications. The preoperative instructions cover the period between the preoperative clinic visit (usually 1 - 2 weeks prior to surgery) up until the day of surgery. The instructions for each medication were: a) continue until surgery, b) discontinue (for a given number of days) prior to surgery, and c) whether to take the medication on the day of the surgery. These instructions were explained to the patient during the consult, written on the anesthetic record, and a copy of those instructions was given to the patient at the end of the consult. On the day of the surgery the nurse who admitted the patient reconciled the patient’s medication list. This reconciliation involved recording for each medication whether the patient took it that day, did not take it, or held the medication for a period of time prior to surgery.
For patients who underwent more than one surgery at HHS, only the first problem-oriented medical record was used and reconciled with the accompanying surgery. Exclusion criteria included patients younger than 18 years of age; patients who underwent emergent surgeries; and patients who underwent living-donor transplants.
Patient identification numbers, date of surgery, and operating room times were collected from the Perioperative Services for HHS. The following patient data were collected from Sovera health records management system: patient medical record number, date of preoperative appointment, date of surgery, planned start time of surgery, planned disposition, surgery, and type of booked anesthetic. The patient demographic data and past medical history that were extracted included: age; sex; ASA physical status class; number of patient medications; medication class for each of their listed medications (Appendix 1); days between preoperative appointment and day of surgery; individual completing medication reconciliation (physician, pharmacist or nurse); patient comorbidities (Appendix 2); surgical class: general, orthopedic, gynecological, cardiac, urological, other (electrophysiological, neurosurgical, otolaryngologic, ophthalmologic, plastic, vascular); medication instructions and compliance with those instructions: Drug taken on day of surgery? (yes/no), Drug not taken on day of surgery? (yes/no), Compliant with preoperative instructions? (yes/no), If not compliant: drug taken on day of surgery when advised that it should be held (yes/no), drug held on day of surgery when advised it should be taken (yes/no), drug not held for the correct number of days (yes/no).
Consequences of non-compliance were assessed by the impact on blood pressure for patients receiving antihypertensives, blood sugar level for patients receiving diabetic medications, drop in hemoglobin for patients receiving anticoagulants, bronchospasm for patients receiving bronchodilators, seizures for patients receiving anticonvulsants, as well as case delays regarding the operating schedule.
Sample Size Justification
The sample size was calculated using a previously identified preoperative medication order non-compliance rate of 22% and the use of 105 investigational predictors (15 patients for each of the seven potential predictors of non-compliance), rounding up to n = 650 to accommodate for missing patient information [5]. Due to incomplete charting, illegible charting, or the lack of medication use, 257 patients were removed, resulting in a final sample size of n = 393. The modified sample size was appropriate based on the objective to determine the relationship between baseline patient demographic and prognostic characteristics and medication non-compliance using regression modelling. To avoid overfitting and produce stable models, we applied the “one in ten” rule of thumb and assigned 10 events per degree of freedom [10] [11]. With the sample size of n = 393 and 146 events of non-compliance, we were able to explore seven continuous and categorical factors of interest with at least 14 degrees of freedom.
Statistical Analysis
Compliance and non-compliance were determined by comparing the medication instructions from the preoperative clinic and reconciliation information from the day of surgery. Patients were identified as non-compliant if any discrepancy was found when comparing the medication instructions from the preoperative clinic and day-of-surgery medication reconciliation. Specifically, non-compliance was assigned for a given medication if the drug was: taken on day of surgery when advised that it should be held, held on day of surgery when advised it should be taken, or not held for the correct number of days.
The overall non-compliance of preoperative medication instructions is reported as a percentage and corresponding 95% confidence intervals (CI). Demographic characteristics are reported by compliant versus non-compliant groups. Continuous variables are reported using group means and standard deviations (SD) and categorical variables are presented as number of patients with percentages. The impact of patient demographic variables (age, sex, ASA class, number of medications, medication class, days between preoperative appointment and day of surgery, individual completing medication reconciliation, patient comorbidities, type of surgery) were evaluated as potential predictors of non-compliance using logistic regression analysis. The impact of non-compliance on intraoperative blood pressure among patients receiving antihypertensive medications was also evaluated using logistic regression analysis. Results are reported using odds ratios and corresponding 95% confidence intervals (CI) and p values, with a statistical significance level set to p ≤ 0.05. Additionally, descriptive statistics were used to present medication instructions given by drug type as well as the relationship between non-compliance and intraoperative adverse events.
3. Results
In total, 393 adult patients undergoing non-emergent surgery met the inclusion criteria within the specified time frame; of these, 146 (37.2%) (95% CI, 32.35% to 41.95%) patients were not compliant to at least one of the medication reconciliation instructions provided by the anesthesiologist at the preoperative clinic appointment. Table 1 describes the patient demographics, surgical characteristics, and preoperative appointment details by patient compliance. The study population, on average, was aged in their late 50 s to early 60 s, had proportionally more women than men, was predominantly ASA class 3, had predominantly orthopedic surgeries, and was seen in the preoperative clinic 10 - 12 days prior to their surgery. The medication reconciliation was most often completed by the pharmacist.
No significant associations were observed for any of the potential risk factors (age, sex, ASA class, number of medications, type of surgery, individual completing medication reconciliation, and time between preoperative appointment to surgery) and non-compliance, in both the unadjusted and adjusted analyses (p > 0.05) (Table 2).
*T-test was used to calculate P value. †American Society of Anesthesiologists physical status classification system. ‡Mann-Whitney U test was used to calculate P value. All non-labelled P values were calculated using Chi-squared tests. IQR, interquartile range.
![]()
Table 2. Relationship between baseline patient data and complete medication compliance.
*Other includes electrophysiological, neurosurgical, otolaryngologic, ophthalmologic, plastic, and vascular surgery types. ASA, American Society of Anesthesiologists.
Overall, patients took a total of 2901 medications before surgery. Patients were instructed to continue taking 891 (30.7%) medications and stop taking 688 (23.7%) medications. There were no instructions given for 1322 (45.6%) of the medications (Figure 1). Medication instructions by drug type are shown in Figure 1. In most patients, anticonvulsants, beta blockers, calcium channel blockers, combination antihypertensives, anti-Parkinson medications, bronchodilators, and thyroid medications were continued up until the surgery; nonsteroidal anti-inflammatory drugs, anticoagulants, and oral hypoglycemics were most commonly withheld. The compliance status of 10 medication types, including analgesics, anticoagulants, anticonvulsants, antihypertensives, anti-Parkinson’s, bronchodilators, diabetic medications, nitrates, thyroid medications, and “other” (vitamins, herbal/natural, antilipemic, etc.) medications is shown in Table 3. ‘Other’ medications include 34 different drug types and are listed in Appendix 3.
![]()
Figure legend: *Includes: alpha 1 blockers, beta blockers, ACE inhibitors and ARBs, calcium (Ca2+) channel blockers, diuretics, and combination antihypertensives; †Other medications include 34 different drug types that are shown in Appendix C. ‡Recommendations based on review article by Muluk et al. in UpToDate and infographic by Wanderer et al. [16] [17] “Hold” refers to continuing therapy up to day of surgery and holding morning dose. “Continue” refers to continuing therapy up to and including day of surgery. §Based on analysis by Roshanov et al. [18]. ACE-I, angiotensin-converting enzyme inhibitors; ARB, angiotensin II receptor blocker; NSAIDs, non-steroidal anti-inflammatory drugs.
Figure 1. Medication instructions by drug type.
![]()
Table 3. Compliance status by drug type.
*Includes: alpha 1 blockers, beta blockers, ACE inhibitors and ARBs, calcium channel blockers, diuretics, and combination antihypertensives. ACE-I, angiotensin-converting enzyme inhibitors; ARB, angiotensin II receptor blocker; NSAIDs, non-steroidal anti-inflammatory drugs.
In terms of consequences of non-compliance including intraoperative hypertension, hypotension, hyperglycemia, bronchospasm, significant bleeding, seizure, and case delay, incidence rates were similar between non-compliant patients and compliant patients (Table 4). The only significant result was seen in patients taking beta blockers for blood pressure control (Table 5). Patients who were told to take their beta blocker on the morning of surgery and did not have a lower mean (SD) peak blood pressure (144.3 [18.4] mmHg) compared to the patients who complied with taking their beta blocker (163 [22.9] mmHg, p = 0.0046). These patients also had a lower incidence of having a peak blood pressure ≥ 160 mmHg (0.2 vs. 0.6, p = 0.0055). There were no significant differences in blood pressure associated with compliance when all antihypertensives were analyzed together, including alpha-1 blockers, beta blockers, angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers, calcium channel blockers, diuretics, and combination antihypertensives.
4. Discussion
Our study showed that among adult patients undergoing non-emergent surgery, the incidence of medication non-compliance was 37.2%. This is consistent with perioperative non-compliance rates documented in the literature, which show a range of 16% - 40% [5] [7] [8].
Age, sex, ASA class, type of surgery, time between preoperative appointments to surgery, and whether the individual who collected the list of patient medications was a physician, nurse, or pharmacist were not found to be statistically significant factors in predicting patient non-compliance. This is in contrast to previous studies, which found significant associations between increased age and higher ASA class [8]. Other risk factors for patient non-compliance in the literature that were
![]()
Table 4. Relationship between medication instruction non-adherence and adverse outcomes.
*Based on difference between preoperative hemoglobin and lowest postoperative hemoglobin. BS, blood sugar; Hb, hemoglobin; SBP, systolic blood pressure.
![]()
Table 5. Impact of patient non-adherence to selected antihypertensive medication instructions on blood pressure.
*T tests were used to calculate P values for continuous measures. Chi-squared tests were used to calculate P values for dichotomous measures. †Includes: alpha 1 blockers, beta blockers, ACE inhibitors and ARBs, calcium channel blockers, diuretics, and combination antihypertensives. ACE-I, angiotensin-converting enzyme inhibitors; ARB, angiotensin II receptor blocker; BP, blood pressure.
not measured in the present study include non-white race, increased number and cost of medications, drug insurance coverage, increased number of physician visits, and decreased number of physicians seen regularly [12]. As this was a retrospective study, the reasons for noncompliance could not be determined.
Surprisingly, there were no consequences of nonadherence to medication instructions regarding antihypertensives, diabetic medications, anticoagulants, bronchodilators, or anticonvulsants. Patients on beta blockers who did not comply with instructions to take their beta blocker on the morning of surgery were found to have a lower mean peak blood pressure (p = 0.0046) as well as a lower incidence of a blood pressure ≥ 160 mmHg (p = 0.0055), compared with patients on beta blockers who complied with the medication instructions. This paradoxical result could have also been a result of the small subgroup sample size (n = 76). Furthermore, when all antihypertensives were grouped together, there was no difference in intraoperative blood pressure. There may be two explanations for why medication non-compliance did not show clinical impact. The first may be that the incidence of some of the adverse events studied such as bronchospasm and intraoperative seizure are very uncommon (0.16% and 0.8%), and our study was not sufficiently powered to detect a signal [13] [14]. Furthermore, many anesthetic agents may have anti-convulsant properties, including isoflurane, barbiturates, propofol, ketamine, and benzodiazepines [15]. The second reason could be that anesthesiologists are routinely able to manage the patient’s blood pressure, blood sugar, and hemoglobin intraoperatively in specified safe ranges, regardless of medications taken preoperatively. Thus, nonadherence with medication instructions may not necessarily translate into intraoperative complications.
A notable finding was that instructions regarding almost one-half (45.6%) of all medications were not reported during the preoperative appointment. This could either mean that verbal instructions were given and subsequently not recorded or forgotten by the patient, or that no instructions were given. One factor that could have contributed to the lack of instructions given is the time pressure anesthesiologists face in the preoperative clinic. At HHS, the preoperative clinic is run by four staff anesthesiologists, a resident, a nurse, and a pharmacist, and sees over 100 patients a day. Future work should study the relationship between the duration of pre-operative appointment and medication compliance.
Another notable finding was that for some drug types, medication instructions given contrasted with the recommendations found in the literature. For example, the literature suggests continuing anticonvulsants, holding angiotensin-converting enzyme inhibitors, and holding diuretics on the day of surgery [16] [17] [18], yet 17.6%, 28%, and 26.7% of the time, respectively, patients were given the opposite instruction (Figure 1). Further research should be conducted examining the reasons behind these discrepant instructions.
Several interventions have been studied in the literature to improve medication compliance. Pfeifer et al. showed that using a standardized electronic medical record-derived medication instruction template, with simplified printed instructions, improved adherence by 6% (p = 0.025) on the day of surgery [7]. Kini et al. conducted a systematic review that identified successful interventions as using combination pills to reduce daily pill burden, clinical pharmacist consultation for disease co-management, and medication-taking reminders such as telephone calls to prompt refills [19]. These strategies can be implemented in a future quality improvement program at our institution.
Our study had several limitations. The first limitation is that the outcome of medication instruction adherence is self-reported by patients; thus, it may be an underestimate. The second limitation is that a signification number of patients (45.6% of medications) who either received no medication instructions or verbal, non-recorded instructions were grouped into the “no instructions given” category, and adherence was unable to be determined. The third limitation is that for the secondary outcomes regarding consequences of non-adherence, the study sample size was not powered to capture the low complication rates of adverse events such as intraoperative hypoglycemia, anemia, bronchospasm, and seizures. Future research examining the relationship between medication instruction nonadherence and adverse intraoperative events requires large sample sizes.
5. Conclusion
In summary, our study shows that 37.2% of adult patients undergoing non-emergent surgery were non-compliant with medication instructions and patients did not receive any written instructions for 46% of their medications. We did not identify any significant risk factors associated with patient non-compliance and non-compliance was mostly not associated with any adverse outcomes.
Acknowledgements
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Author Contributions
James E. Paul: conception and design, analysis and interpretation of data, drafting the article, revising the article critically for important intellectual content, final approval of the version to be published.
Henry He: analysis and interpretation of data, drafting the article, revising the article critically for important intellectual content, final approval of the version to be published.
Justin Diep: drafting the article, final approval of the version to be published.
Thuva Vanniyasingam: analysis and interpretation of data.
Sean Middleton: conception and design, acquisition of data, analysis and interpretation of data, final approval of the version to be published.
Abbreviations
ASA, American Society of Anesthesiologists
CI, Confidence interval
HHS, Hamilton Health Sciences
SD, Standard deviation
Appendix
Appendix 1. Medication Classes
5-ASA, 5-aminosalicylic acid; ACE, angiotensin-converting enzyme; CNS, central nervous system; NSAID, non-steroidal anti-inflammatory drug.
Appendix 2. Patient Comorbidities
COPD, chronic obstructive pulmonary disease; LV, left ventricle; GERD, gastroesophageal reflux disease; TIA, transient ischemic attack.
Appendix 3. “Other” Medications
CNS, central nervous system.