Pakistan comprises four provincial localities, Punjab, Baluchistan, Sindh, and Khyber Pakhtunkhwa, and it is also a land of two independent administrative territories, Gilgit Baltistan and Azad Jammu Kashmir. Provinces are subdivided into divisions, districts, tehsils (administrative areas that consist of towns), and towns. Though the land occupied by Punjab is only 26% of Pakistan's total area, it is the most populous of all provinces, home to 60% of the country’s population (Pakistan Bureau of Statistics, 2021). The government offers healthcare through a three-tiered healthcare delivery system, which provides coverage for approximately 10% of the population. Basic health units and rural health centers provide the foundation of primary healthcare. Secondary care is delivered by tehsil and district headquarters, which encompass both first and second-referral facilities, as well as tertiary care supplied by teaching institutes. The private health sector incorporates an extensive array of medical professionals, including physicians, nurses, pharmacists, traditional healers, and laboratory technicians (Wikipedia, 2023). The physician-to-patient ratio in Pakistan is 1.1 for 1,000 patients, which is much lower than the standard values, and most of them are engaged in private practices (Economics, 2023).

This multisite study was performed using the cross-sectional study design. We have gathered information from the physicians working in government hospitals and private hospitals. These locations were selected through convenience sampling, but they carry a limitation in that such sampling lacks proportionate representation of the entire population, resulting in lower generalization of research findings than probability sampling.

Our target population was practicing physicians in the government and private hospitals from the three cities (Lahore, Sahiwal, and Bahawalpur) of the Punjab, Pakistan. Data were collected by three teams of data collectors (mostly students) who were trained to collect the data according to the protocol. These hospitals provided both inpatient and outpatient services for the population. We targeted the physicians providing services in inpatient wards and outdoor clinics. The included specialties were general practitioners, obstetricians and gynecologists, oncologists, surgeons, dentists, and orthopedists. Physicians working in academia, on leave, or holding administrative posts in hospitals were not included.

The study was conducted from October 2023 to April 2024. Initially, we approached the potentially eligible physicians in hospitals to ascertain their willingness to participate in the study. We extended invitations to nearly all the accessible physicians during the time of the visit, and those who expressed their willingness and availability were provided with the questionnaire to complete on the spot. The primary issues encountered were a high patient load and limited time availability of the doctor. To address this challenge, we approached the physicians during their break and provided them with the questionnaire. We subsequently gathered the completed forms after some time. Even after this, if we faced a lack of success, we approached the physicians as patients and politely asked if they would be willing to provide the data. Physicians who remained unwilling to participate were excluded. Prior to the initiation of data collection, a cover letter was provided to participants who agreed to participate. The cover letter communicated to physicians the need to maintain anonymity in their responses, emphasized the voluntary nature of their involvement, and provided physicians with an understanding of the objective of our study. Physicians were asked to sign the letter indicating that they approved of the participation. We asked the participants to complete the questionnaire on the spot, but if they were not free to complete it, they could return the completed questionnaire via mail or deliver it to a team member visiting to collect. We assured the participants that this survey was for research purposes and that the confidentiality of their responses would be maintained to ensure the appropriateness of the data. We also explained that the study was conducted by students of Xi’an Jiaotong University, China, to improve the standards of opioid prescribing, strengthening the policies to rationalize the use and aligning the healthcare professionals’ role in managing the opioid crisis in Pakistan. A questionnaire was considered useful if it was filled out and provided a single answer for questions requiring one response, and so on. To carry out the thorough examination of data completeness, the principal investigator independently verified and entered the data in MS Excel, which was cross-checked by AHG.

A self-administered, structured questionnaire comprising five sections and 64 response items that spanned four pages was distributed to respondents in our study. This survey instrument was developed based on a study of relevant literature (Majid et al., 2019; Wenghofer et al., 2011; Morley-Forster et al., 2003; Hutchinson et al., 2007; Nwokeji et al., 2007), which we further improved through discussions between the authors (MA, HA, and YF). The fundamental use of this form is the reporting of opioid-related adverse events (e.g., by number of patients, type of opioid, and cause of the event) and physicians' concerns regarding the different effects of opioid prescribing on patients. We also measured physicians’ comfort, confidence, and satisfaction with opioid prescribing; their views on approaches to adjust their prescribing; and their interactions with other healthcare professionals.

The first section of the questionnaire had information regarding gender, years of practice, age, medical specialty, practice location, and type of hospital. The second component consisted of 11 questions pertaining to the understanding of opioids, opioid stewardship, and the resolution of opioid-related issues. The responses were recorded in binary format (yes/no), and by selecting the appropriate answer, the correct response was collected. Section 3 of the questionnaire consisted of seven questions that examined attitudes and 15 questions that examined prescribing practices related to opioids. The responses were collected using a 5-point Likert scale for prescribing attitudes. This section included an additional 15 questions about opioid prescribing practices. These questions inquired about the number of patients who were administered opioids, the reasons for not prescribing opioids if none were provided, the reporting of adverse events related to opioid usage, and the factors of opioid adverse events. The questions were based on attitudinal queries frequently utilized in surveys regarding physician attitudes and practices (Hutchinson et al., 2007; Nwokeji et al., 2007). Explorations suggest that these beliefs are linked to physicians’ tendency to prescribe opioids (Wenghofer et al., 2011; Morley-Forster et al., 2003; Hutchinson et al., 2007; Nwokeji et al., 2007). Seven items were present in Section 4 of the questionnaire, which asked about physicians' concerns when prescribing opioids, and a 4-point scale evaluated the section (not at all concerned, a little concerned, somewhat concerned, and very concerned). The last section included information on the interaction of physicians with other health care providers (HCPs) (17 items: frequently, sometimes, and not at all). The questionnaire is available as Supplementary Material S1.

Before commencing the research, the tool was piloted with 11 general medicine physicians, six gynecologists, and six orthopedists to evaluate the lucidity and comprehensibility of the tool items. Pre-testing also helped us determine whether respondents were willing to participate in the study and provide the necessary information. Twenty-three physicians participated in the pretest survey, and their responses were not included in the original data. Based on their feedback, minor alterations were made to clarify the wording and scaling.

The construct and content of the questionnaire were validated by an extensive literature review, ensuring that the data collection instrument accurately measured the intended hypothesis. Professional researchers examined the items in the data collection instrument and provided their professional comments to guarantee face validity. The designed questionnaire underwent a review and evaluation by four professors with a background in pharmacy. Their aim was to analyze the suitability of each item for measurement purposes. The instrument was revised based on the expert comments obtained through discussions with researchers. Internal consistency of the questionnaire was measured by Cronbach’s alpha. Cronbach’s alpha of the questionnaire was 0.669≈0.70, which shows that the reliability of the research tool is acceptable.

An analysis of descriptive statistics was conducted to examine the demographics and other variables. Percentages were used to represent the ordinal data, while the mean and standard deviation were used for continuous variables. The responses to each item in the knowledge, attitude, and concerns section were recorded and are represented as numerical values and percentages. We employed Spearman’s rho correlation test to assess the relationship between the knowledge, attitude, and concern sections. The significance threshold was set at 0.001. To assess the correlation, we transformed the knowledge (items 1–11) into scores by assigning a value of 1 to the correct answer and 0 to the incorrect response. In order to obtain the scores for the attitude portion (items 3–9), we aggregated the responses, assigning a value of 5 to “strongly agree,” 4 to “agree,” 3 to “neutral,” 2 to “Disagree,” and 1 to “Strongly disagree.” Similarly, the concern section (items 2–9) was transformed into scores by assigning a numerical value to each response. For instance, a response indicating “very concerned” was assigned a score of 4, “somewhat concerned” was assigned a score of 3, “a little concern” was assigned a score of 2, and “no concern at all” was assigned a score of 1. The scores were aggregated and utilized in correlation.

The study received ethical approval from the Bioethics Committee of Xi’an Jiaotong University, China (2023-21-PA). Prior to beginning, individual physicians were mandated to submit written consents. The researchers abstained from soliciting any personal information from the individuals. In addition, every participant received an assurance that their data would exclusively be utilized for research purposes and would remain anonymous.

Of the 1,009 physicians approached for the study, 816 (80.9%) responded to the study and completed the questionnaire. Their age ranged from a minimum of 25 to a maximum of 55 years, with a mean age of 38 years. The practice years ranged between 1 year and 30 years, and the mean was 16 years. More men (58.8%) participated than women (41.2%), and similar percentages of respondents worked at government hospitals (56.9%) vs. private settings (43.1%). The largest group of participants was general practitioners (47.1%), followed by surgeons (19.6%). All the data are available in Table 1.

Regarding knowledge about opioids and opioid stewardship, respondents are highly aware of most of the questions. The “yes” responses vary from a minimum of 51.0% (indicating awareness that opioids are classified under the G schedule of the drug regulatory authority of Pakistan) to a maximum of 96.1% (indicating knowledge of opioid drugs). Remarkably, the participants exhibited a high level of awareness regarding the potential hazards and negative consequences linked to opioid therapy (92.2%) as well as the factors that increase the likelihood of opioid abuse and addiction (90.2%). When rating the opioid items, the maximum score is 11. In our study, the participants were aware of a maximum of 11 questions related to opioids, while the minimum number of questions answered was 2. The mean (standard deviation) of the score was 8.14 (2.37), indicating that individuals had a good understanding of opioids (Table 2).

The percentage of individuals who expressed positive attitudes and confidence toward opioid prescribing for chronic pain increased significantly from 5.9% to 64.7% when considering concerns about potential opioid diversion or abuse in their practice setting. A notable finding was that a mere 19.6% of participants, or 1 in 5, reported feeling satisfied when prescribing opioids for pain patients in their respective settings (the details are presented in Table 3). The items were rated on a 5-point Likert scale, with a rating of 5 indicating strong agreement and a rating of 1 indicating severe disagreement. The score spans from 7 to 35. Of all the participants, 16 individuals (2% of the total) had the lowest attitude and scored 7. On the other hand, 32 respondents (3.9% of the total) achieved the highest score of 29.

Upon examining the prescribing practices, we discovered that over half (51.0%) of the physicians never prescribed opioids in their practice, whereas only 5.9% prescribed opioids to more than 200 patients, 32 (3.9%) physicians prescribed opioids to 1–100 patients, and 32 (3.9%) physicians prescribed opioids to 100–200 patients in past 3 months. Figure 1 illustrates the factors that contribute to the decision not to prescribe opioids. Of the entire sample, 31.3% of the physicians had experienced an adverse opioid drug event in their practice in the past 3 months. Figure 2 illustrates the predominant opioids responsible for the occurrence of adverse events.

Physicians reported morphine was involved in 31.4% of adverse events during their prescribing practices. Participants also highlighted contributing factors of adverse events due to opioids in their practices: 39.2% of physicians reported that the prescribed opioid dose was too high during the occurrence of adverse events, whereas 43.1% of adverse events were reported because the patient took more opioids than were prescribed. All other factors responsible for adverse events due to opioids are listed below in Figure 3.

Nearly one in five (19.6%) of the prescribers expressed indifference to patients' drug addiction and non-compliance with therapy. Likewise, a minor fraction (11.8%) had no concerns regarding issues with legislation and expectations from patients. The specific information is provided in Table 4.

Fewer than half (41.2%) of the physicians surveyed reported that they had never engaged with pharmacists regarding opioid therapy and prescriptions. Conversely, only 21.6% indicated that they had never engaged in any form of communication with nurses regarding opioid therapy. The primary issue found when contacting a pharmacist is that the pharmacist appeared to lack expertise on opioids, with a prevalence of 69.6%. This was followed by difficulty reaching directly due to communication gaps, accounting for 66.7% of the cases (Table 5). Most respondents (64.7%) reported that pharmacists effectively addressed their inquiries or concerns regarding opioids. Most respondents had a positive perception of nurse-physician interactions. Specifically, 76.5% agreed that nurses accurately assessed opioid intoxication, withdrawal, or pain. Additionally, 70.6% of respondents reported that nurses occasionally or frequently encouraged them to prescribe benzodiazepines or analgesics to keep patients calm or quiet (Table 5).

Table 6 presents the results of demographic associations of the respondents with the knowledge, attitude, and concerns scores. Analysis revealed that men had statistically high knowledge and practice scores, whereas women had higher concern scores. Respondents working at private hospitals scored statistically significantly higher than respondents working at government hospitals (P < 0.001).

When we examined the relationship between knowledge, attitude, and concerns, we found a substantial positive connection between the knowledge score and attitude score (r = 0.275, p = 0.01), the knowledge score and concerns score (r = 0.120, p = 0.01), and the attitude score and concerns score (r = 0.495, p = 0.01). The details are given in Table 7.

We recognize the constraints of our study. This study inquired about chronic pain, encompassing both cancer-related pain and palliative pain. It is important to note that distinct opioid prescribing rules are applicable to palliative patients. This study was limited to a cross-sectional snapshot and did not inquire about the respondents' interest in further exploring specific clinical subjects, such as the utilization of addiction screening instruments. These aforementioned concerns can be resolved through future studies that aim to confirm the dependability of perceived levels of confidence in opioid prescribing by comparing them to actual practice patterns. Conducting a qualitative study will be beneficial in thoroughly understanding the viewpoints of physicians regarding their worries and the adverse occurrences experienced by patients.