Introduction

The 11th revision of the International Classification of Diseases (ICD-11) included burnout as an occupational syndrome adversely affecting health status or healthcare services and arising from unmanaged prolonged work-related stress1. Three characteristics define burnout: “feelings of energy depletion or exhaustion; increased mental distance from one’s job, or feelings of negativism or cynicism related to one’s job; and reduced professional efficacy”1.

Physicians faced a 41% higher risk of burnout than the general population2. Burnout was recognized as a global healthcare problem since it affected more than 50% of clinicians in the United States, with similar or higher prevalence recorded in other countries3. In Egypt, burnout prevalence ranged from 22 to 77% among physicians, with a pooled rate of 60.59%4. Front-line specialties such as emergency medicine and family medicine are more prone to burnout5.

Burnout was linked to workplace stressors, including electronic medical records, long and conflicting work hours, insufficient compensation, work-life imbalance, excessive work demands, unsupportive coworkers and leadership, and limited control over the work environment6. Disparities in burnout susceptibility among physicians may be attributed to personal traits such as ineffective coping mechanisms, doubt, perfectionism, neuroticism, young age, relationship status, female gender, partner occupation, and immature children7. Despite not being categorized as a medical condition, burnout has gained much attention in recent years due to its multifaceted consequences on the patient and physicians themselves6.

A primary cause of avoidable injury in health care is drug safety incidents8. Medical errors ranked as the third most frequent cause of death in the United States9. Among healthcare recipients, 1 out of 30 suffered an avoidable medication-related injury; 26% of these are classified as serious or life-threatening, and 58% occurred during the prescribing stage10. Prescription errors were the most frequently reported medication errors in Egypt (54%)11. An estimated annual cost of US$ 528.4 billion was related to suboptimal medicine use-related morbidity and mortality in the U.S., representing up to 16% of all healthcare costs in the country in 201612.

Despite the cumulative data that links burnout with adverse patient outcomes, current research has primarily relied on physician and patient self-reports of errors rather than objective measures of patient outcomes13,14. In a systematic review focusing on physician burnout and its impact on patient outcomes, an analysis of 14 studies assessing physicians’ self-perceptions revealed a significant association between burnout and suboptimal care or medical errors. However, an objective evaluation of clinical charts via 5 studies failed to detect any relationships15. Due to self-criticism that can be exaggerated by burnout, physicians’ self-reported data on their performance is less reliable16. In a cross-sectional study including 7395 surgical residents in the United States, burnt-out residents were three times more likely to self-report harmful medical errors; however, there was no difference in objectively evaluated postoperative outcomes for surgical patients17. Additionally, depression was linked to both physician burnout and medical errors, except for a few studies that reported the association of objectively measured medical errors with depression in contrast to burnout18,19,20,21.

In 2019, the World Health Organization (WHO) reported minimal progress in the global reduction of patient harm over the past 15 years22. They recommended a better understanding of intangible aspects influencing patient safety and their implication in policy and practice, particularly in primary care, where half of the global burden of patient harm occurs22. Over half (53%) of the avoidable medication-related harm worldwide occurred at the prescribing phase, and the percentage increased to 80% in low- and middle-income countries23. Because physicians’ burnout influences healthcare quality, it’s critical to comprehend how burnout affects objectively measured patient outcomes. Therefore, this study aimed to investigate the prevalence of burnout and medication prescribing errors among primary healthcare physicians and the association between them after adjusting for depression.

Methods

Study design and participants

This was a multi-centric cross-sectional study of family physicians in 25 primary health care (PHC) units in Alexandria between March 1, 2022, and February 28, 2023. Alexandria governorate takes up the ninth rank in descending population distribution of Egyptian governorates in 202424. In Egypt, Family Health Units (FHU) are the initial stage of comprehensive care for the person, family, and community25. In 2021, family health units represented 90.5% of primary health care settings in Alexandria26. FHU has a specific list of 110 essential medications and vaccines.

All participants signed written informed consent. The research was approved by the ethical review board of the Medical Research Institute and the Ministry of Health and Population, and it adhered to the Declaration of Helsinki’s ethical guidelines. All the PHC clinicians in the family care units at the time of data collection were eligible for voluntary participation.

Sample size

According to a pilot of available registers of family physicians, the total target population was approximately 500 family physicians working in Alexandria PHC units. At a 95% confidence interval and 5% precision limit, a minimum sample size of 181 physicians was calculated based on the estimated burnout prevalence of 24.3% from a previous research in a Saudi Arabian primary care setting27. For the second outcome, the proportion of burnout in physicians who reported errors was 77.6% versus 51.5% in those who did not report errors28. A sample size of 118 participants can confirm this effect with 80% power and a 5% significance level by using Epi Info 729.

Study measures

The physicians received a self-administered survey categorized into 3 categories: participants’ characteristics, the Professional Fulfillment Index (PFI), and the Patient Health Questionnaire-9 (PHQ-9) to assess burnout and depression30,31. We employed the Questionnaires’ English version.

Participants’ characteristics

Basic demographics included age, sex, marital status, number of children, partner work, ride time to work, exercise frequency, history of anxiety or depression, and chronic diseases. Participants’ occupational details included practice years, certificates, occupation, employer status, practice setting, managerial status, number of patients seen per shift, weekend work frequency, and intention to quit the practice.

Burnout and professional fulfillment measurement

The validated 16-item PFI survey uses a 5-point Likert scale to assess professional fulfillment (6 items) and the overall burnout scale (10 items)31. Response options for professional fulfillment items varied from “not at all true” to "completely true," while burnout scale alternatives ranged from “not at all” to “extremely”31. Permission was acquired for using the questionnaire.

The scale scores are derived by averaging all items’ scores on each related scale and then multiplying by 2.5 to scale from 0 to 10. Professional fulfillment and overall burnout scale cut points are 7.5 and 3.325, respectively32. Higher scores indicate favorable outcomes in professional fulfillment and negative experiences in burnout32.

Depression measurement

The PHQ-9 is a validated, self-administered tool for identifying, categorizing, and monitoring depressive symptoms30. The frequency of depressed symptoms was evaluated during the preceding two weeks using four responses ranging from 0 "not at all" to 3 "nearly every day", with a total score between 0 and 2730. A score of 10 or higher was considered indicative of depressive symptoms in participants33.

Medication prescribing errors

A prescribing error occurs when the physician’s ordering decision or writing process results in a lower probability of efficacy or a higher risk of adverse effects of treatment34. Based on the Ghaleb et al. and Dean et al. scenarios, the prescribing errors were classified with adjustments to local clinical reporting forms and the study’s outpatient setting:

  1. 1.

    Decision-ordering error options included drugs with no valid indication, therapeutic duplication, indication without medication, drug contraindication, drug interaction without proper management, wrong dosing regimen, wrong duration, wrong dosage form, or wrong dilution instruction34,35.

A wrong dose was identified as exceeding 20% of the recommended dose for a drug of wide therapeutic index, and prescribing when needed (PRN) without specifying the minimal dose interval36,37.

  1. 2.

    Process ordering error options included the wrong patient, incomplete or wrong required patient information, wrong drug name, illegible handwriting, wrong strength, wrong decimal point, incomplete prescribing instructions, repetition, or unapproved abbreviations34,35.

The patient’s name, age, gender, diagnosis, and child weight (if appropriate) are the basic required patient information in the prescription36.

Medication prescribing errors detection process

In Egyptian PHC, prescriptions are handwritten without adopting standard coding for diseases or electronic prescribing systems. The Prescription has 1 or more prescribed medications. Researchers collected the medication data by reviewing the first 50 available prescriptions for each physician in the primary health care units’ pharmacies. Two experienced clinical pharmacists independently checked the appropriateness of drug therapy for the patient’s condition according to best practice guidelines.

Evidence-based sources included Lexicomp, UpToDate, British National Formulary (BNF), Centers for Disease Control (CDC), National Institute for Healthcare and Excellence (NICE), Drugs.com, Medscape, disease-specific guidelines, and others for identifying ordering errors.

The pharmacists recorded a description of the event, the drug’s category, the classification of ordering error types, and the categorization of the potential harm of each event. The classification of the potential harm associated with ordering errors varied from none to the potential of severe harm38. The pharmacist reviewers’ disagreement was settled by consensus via discussion. The drugs were classified using the second level of the WHO Anatomical Therapeutic Chemical (ATC) categorization system39.

Statistical analyses

The ordering error rate (%) was obtained by dividing total errors by the number of prescribed medications and then multiplying by 10040. Multiple errors in the same medication were counted separately.

Quantitative data were presented as a median plus interquartile range, while qualitative data were summarized as frequencies and percentages. The Kolmogorov–Smirnov test detected the continuous data’s deviation from normality41. Due to the overdispersion in the poisson models, negative binomial analysis was used to identify parameters related to prescribing error rates, with the number of prescribed items as an offset in the models. The multivariate regression model included all variables with a P value <0.05 in the bivariate analysis. Data were analyzed using Statistical Package for the Social Sciences (SPSS version 28) and R Software, with a significance level set at p < 0.0542,43.

Results

Characteristics of the respondents

The study included 184 family physicians (Fig. 1). The median age of the participants was 45 (interquartile range [IQR] 38.00–55.00) (Table 1). Most participants were females (94.02%), married (81.52%), and parents (86.41%). Less than half of the participants (44.57%) had at least one comorbidity, while 19.02% of the physicians reported a prior anxiety or depression diagnosis. Two-thirds (71.74%) of the respondents were sedentary, and only 6.52% exercised daily.

Most responders were family physicians (66.85%), full-time employees (93.48%), and non-managers (83.15%), with a median of 20 practice years (IQR 11.00–30.00). About one-quarter of respondents had only a bachelor’s degree, 36.41% completed a postgraduate diploma, and 40.22% earned master’s or higher degrees. The median number of patients examined per shift was 30, with an IQR of 20–40.

Fig. 1
figure 1

Flowchart of data collection and analysis.

Full size image
Table 1 Characteristics of the participants in primary health care centers in Alexandria, Egypt, 2022–2023 (n = 184).
Full size table

Physicians’ thoughts of abandoning the profession

In the previous 12 months, over 1 in every 3 participants frequently considered leaving their profession (34.78%), 33.15% of respondents thought about abandoning their profession occasionally, and 32.07% rarely or never.

Prevalence of burnout, professional fulfillment, and depression in the study’s sample

About half of the participants were depression-symptom-free (57.07%), while 78 physicians (42.39%) met the criteria for burnout, and 164 (89.13%) had a low personal professional fulfillment score (Fig. 2).

Fig. 2
figure 2

Prevalence of burnout, professional fulfillment, and depression among the study’s participants in primary health care centers in Alexandria, Egypt, 2022–2023.

Full size image

Types and prevalence of prescribing errors

A total of 5900 prescriptions for 118 family physicians were analyzed. The review of 14,121 prescribed medications revealed 6543 prescribing errors (PE) (46.34%). Errors in the prescribing decision-making occurred more frequently than in the process (63.12% vs. 36.88%). The wrong dose regimen was the most prevalent type of error (53.68%), followed by incomplete patient details (24.35%) and incomplete prescribing instructions (12.52%) (Table 2).

Table 2 The types and frequency (%) of prescribing errors at family health units in Alexandria, Egypt, 2022–2023.
Full size table

The percentages represent the proportion of the total number of prescribing errors.

Potential Harm of Prescribing Errors

Half of the prescribed errors posed minor potential harm to patients, 33.46% moderate severity, and 13.79% no potential risks. Only 0.11% of errors were serious, and there was no potential for severe harm. The category of ordering errors was associated with the potential severity of errors (P ˂0.001) (Fig. 3). Decision and process ordering errors were primarily responsible for minor potential harm (60.46% vs. 39.29%), and 37.6% of decision prescribing errors were moderate in severity, compared to 26.32% of process errors. Only 0.17% of decision errors had a potentially serious impact, while process errors did not contribute to serious concerns. Examples of prescribing errors and potential harm are provided in Supplementary Table S1.

Fig. 3
figure 3

Potential severity of prescribing errors based on the error types in primary health care units, Alexandria, Egypt, 2022–2023. ME, medication error.

Full size image

The top 10 misprescribed drug categories

Based on the ATC classification system, Fig. 4 illustrates the top ten medication categories with prescribing errors. Analgesics, systematic antihistaminics, and systemic antibacterials were the most frequently misprescribed drug classes (20.01%, 17.65%, and 17.15%, respectively). Supplementary Table S2 indicates drug classification according to the ATC system.

Fig. 4
figure 4

Top drug classes involved in the prescribing errors according to the anatomical therapeutic chemical classification system in primary health care units, Alexandria, Egypt, 2022–2023.

Full size image

Factors associated with prescribing error rate

Table 3 illustrates the findings of the unadjusted and adjusted analysis of variables linked with the prescribing error rate. In the unadjusted analysis, the factors with a significant association included gender, chronic illness history, occupation, practice location, and occasional thoughts of leaving the profession.

Table 3 Negative binomial regression analysis of Participants’ Characteristics related to ordering error rates per 100 prescribed Items.
Full size table

Only being male, occupation specialist, and working in rural settings were significantly associated with the error rates in the multivariate analysis (Fig. 5). The prescribing error rate for male physicians was 1.5 times greater than for female ones (95% CI 1.13–2.01; p = 0.005). Significantly higher error rates were linked to specialist occupation, compared to a family physician position with RR 1.29 (95% CI 1.09–1.54; p = 0.003). Rural employees had a 20% higher risk of medication errors than urban workers (RR, 1.20, 95% CI 1.03–1.40; p = 0.025). Depression symptoms were not associated with a higher risk of medication error rate (RR, 1.03, 95% CI, 0.87–1.21; p = 0.751). Similarly, burnout, low professional fulfillment, and quitting thoughts were not significant risk factors for prescribing errors.

Fig. 5
figure 5

Factors related to prescribing errors in primary care settings in Alexandria, Egypt, 2022–2023: a Multivariate Negative Binomial Analysis. RR, rate ratio; CI, confidence interval. *Statistically significant associations with a p-value < 0.05. † Reference Group.

Full size image

Discussion

This multicentric study of family physicians identified that most participants had low professional fulfillment, while almost half of them reported depression and burnout symptoms. The prescribing error rate was 46.34%, but only 0.11% represented serious potential patient harm. The most prevalent ordering errors were incorrect dosage regimens, incomplete patient data, and uncompleted prescribing information. No significant association was detected between depression, burnout, and low professional fulfillment and objectively evaluated prescribing errors. Significant risk factors for prescribing errors included male gender, specialist occupation, and rural work settings.

High burnout prevalence in the current study confirmed the alarming rates among primary care physicians, comparable levels in the United States (35%) and Japan (34%) emphasize the global burden of the problem44,45. Although lower burnout rates were reported in Saudi Arabia (25.2%) and Qatar (12.6%)46,47, The heterogeneity in the measurement instruments and burnout criteria might explain the variability among the studies, especially with the high frequency in burnout subcategories48. The staff shortage, elderly population, elevated medical demands, challenges in providing appropriate patient care, and financial constraints increase primary care physicians’ risk of burnout49.

Professional fulfillment was inversely linked to physicians’ quitting intentions50. Turnover of physicians entails productivity loss, recruitment, and training costs for organizations51. Therefore, the low rate of professional fulfillment in the current study raises a serious alarm, particularly as it aligns with earlier research among Egyptian physicians52,53. Similarly, a survey of 211 primary care doctors in Bahrain revealed an 82.6% low personal achievement rate54.

Depression is one of the prevalent mental illnesses among clinicians55. There is conflicting evidence regarding the association between burnout and depression in physicians56. The current findings highlighted the comparable rates between the two phenomena in the sample. depression prevalence in the current study lines up with the national research of 15,243 emergency doctors in China, in which 35.59% of the participants suffered from depression57. A meta-analysis study among physicians concluded a lower pooled global prevalence of depression (20.5%), but the study was limited by moderate bias risk and heterogeneity across included studies58.

Medication errors are responsible for more than 50% of all preventable injuries in health care worldwide, with an annual cost of 4.5–21.8 billion euros in Europe10. Prescribing errors occur frequently in outpatient and ambulatory healthcare facilities59. The present study’s ordering error rate supported the previous finding in different countries as a Bahraini primary healthcare department (44.3% of the prescribed drugs) and Malaysian primary care clinics (48.0% of prescriptions)60,61. Misagh et al. reported prescribing errors in each of the studied 2000 outpatient prescriptions in Iran62.

Literature studying prescribing errors shows variability in definitions, rate measuring methods, handwritten or electronic prescriptions, investigated population, and evaluation methods63,64. Medication error rate is influenced by the denominator of measurement, such as total prescription, prescribed item, or patient65. The global range of prescribing errors in outpatient settings was 0–91% of the total number of prescribed medications59. Investigations from Bahrain, China, and Saudi Arabia revealed a lower prescribing error rate (34.1% of prescriptions, 4.66% of prescriptions, and 0.09% of prescribed items respectively)66,67,68.

Most errors did not seriously threaten patients. In general practices in England, Avery et al. reported severe errors in 0.18% of the studied prescriptions69. Compared to actually measured harm, a recent systematic review also reported that the potential harm of medication errors in primary care is overestimated since they rarely cause emergency department visits or hospitalizations70.

The most prevalent prescribing error type in our study was the wrong dosage regimen. It was consistent with Abanmy et al.'s findings that incorrect doses accounted for 57.2% of prescribing errors in a Saudi family medicine setting67. Our results aligned with previous systematic reviews in outpatient settings and hospitals59,71.

Incomplete drug orders could jeopardize the quality of patient care by causing time loss, prescription misinterpretation, and improper medicine supply and administration72. Over 36% of prescribing error rates in the current study were writing process errors. Similarly, a high prevalence of incomplete information was observed in the Saudi hospital settings (25%) as well as the English general practice (26.1%)73,74. However, the current study supports previous findings that decision errors are more likely to threaten patient health compared to writing process errors75.

Electronic prescribing systems provided a successful solution to illegible handwriting and incomplete prescription information76. Despite links between electronic prescribing and fewer medication errors and adverse drug events, better patient safety, and time and cost savings, several concerns were raised about e-prescribing adoption. Electronic prescribing systems’ challenges include novel types of errors, implementation and maintenance expenses, team training, healthcare workers’ additional duties, prescribing software systems’ variability, and technical problems77. In addition, pharmacist intervention and physician education can help minimize prescribing errors78.

The current study’s top 3 inappropriately prescribed categories were consistent with the most prescribed drugs in outpatient settings in both India and Nepal79,80. Comparable results, in a different order and frequency, were observed in an investigation in Jordan, in which antibiotics accounted for 31.7% of prescribing errors, followed by analgesics (26.7%), and antihistamines (10.9%)81. However, the investigated family units used a local essential drug list, making it difficult to compare the specific drug data with prior literature.

Depression has a well-established association with burnout, whereas previous studies detected depressive symptoms as a significant predictor of medical errors independent of burnout19,20,82. Compared to non-depressed residents, depressed physicians committed significantly higher rates of medication errors on active surveillance83. In contrast, the rate of total medication error was unrelated to depression symptoms in the current study. A cohort study of 537 residents can explain this disparity since depression showed a significant association with harmful errors but not total medical errors18.

Using objective measures of prescribing errors, the current findings did not support the established association between physicians’ burnout symptoms and self-perceived medical errors in the literature even after controlling the depression presence84. The present findings are consistent with research that evaluated burnout’s association with physicians’ performance using medical charts and orders. Hewitt et al. reported no significant associations between residents’ mental health and objectively evaluated surgical outcomes, despite that burned-out residents are nearly 3 times more likely to self-report harmful medical errors17. Burnout did not significantly impact medical errors collected by independent research assistants among 31 French intensive care units19. Using active surveillance, Fahrenkopf et al. identified no significant difference in the rate of medication error between burnt-out and non-burnt-out residents83. According to a cohort study among American and Canadian resident physicians, burnout symptoms were unrelated to objectively detected medical errors18.

The inconsistent association between burnout and self-reported and observed medical errors may be attributed to physicians’ imprecise self-assessment of their performance85. Burned-out medical professionals are more prone to self-criticism, and hence more likely to disclose errors16. In addition, self-reported data are susceptible to social desirability and recall bias86.

Our findings indicated that errors were more likely to occur with male physicians. These outcomes were consistent with Assiri et al.'s cohort research in Saudi family medicine facilities. The research indicated that male physicians had a 1.6 times higher risk of committing PE than female physicians87. female clinicians were more likely to deliver preventative care, counsel patients, and comply with clinical guidelines than their male colleagues88.

Compared to consultants of different specialties, family physicians had a significantly lower rate of prescribing errors in the current study. Jayaweera et al. also indicated that general internists were more likely than family physicians to prescribe potentially inappropriate drugs89. Although the previous finding lacks a clear explanation, Tonelli et al.'s finding of significant variance in the complexity of patients examined by various medical professions may provide some insight. They identified family physicians’ patients as having the least complexity out of 13 medical specialties90.

According to the present study, there was a significant association between rural work environments and prescribing errors. Research in ambulatory care settings has also reported that rurality increases the risk of inappropriate prescriptions of antibiotics and geriatric medications91,92. With a growing elderly population and health demands coupled with a shrinking health system efficiency and resources, WHO identified primary care disparities among rural and urban regions as a global health threat93.

To the best of our knowledge, this study could be considered the first to investigate the association between objectively measured prescribing errors and burnout among primary care physicians in the region. A further strength of the study was its multicenter design, which involved 25 primary healthcare units. Lastly, the researchers who investigated prescribing errors were unaware of the participants’ levels of burnout and depression, and the study’s specific hypothesis was not disclosed to the participants, so there was no risk of change in the prescribing pattern.

On the other hand, certain limitations need to be considered when interpreting the study results. Firstly, a cross-sectional study design cannot establish causality. Secondly, the study reviewed only essential drug list medications in family health units, which may be misleading due to their limited variety. Thirdly, while two qualified pharmacists identified potential harm from errors, patients were not monitored for actual injuries. Fourthly, prescription errors lack data on inter-rater reliability. Fifthly, the analysis included only physicians who voluntarily agreed to participate in the study, and the majority of participants were females. Although this reflects the current situation in primary healthcare units participating in the study, the nonrandom sample may limit the findings’ generalizability.

Conclusion

The alarming rates of burnout and depression in doctors highlight the necessity of understanding and handling these problems. Although prescribing errors were widespread in primary care settings, physician burnout and depression weren’t linked to them. Significant risk factors for prescribing errors were identified as male physicians, rural work locations, and non-family physician specialists. Longitudinal studies are necessary to evaluate the impact of physician burnout on objective measures of patient safety and enhance prescribing practices in ambulatory care settings.