Environmentally sustainable anaesthesia: a cross-sectional survey of anesthesiologists in teaching hospitals of Karachi, Pakistan

Abstract

Background

Anaesthesia is a recognized contributor to healthcare-related greenhouse gas emissions and operating room waste. While high-income countries (HICs) have begun implementing sustainability initiatives, evidence from lower-middle-income countries (LMICs) is limited. This study aimed to assess the knowledge, attitudes, and practices of anesthesiologists in teaching hospitals in Karachi, Pakistan, regarding environmentally sustainable anaesthesia, using the World Federation of Societies of Anesthesiologists (WFSA) consensus principles as a benchmark.

Methods

We conducted a cross-sectional survey of consultant anaesthesiologists, trainees, and staff anaesthesia medical officers in College of Physicians and Surgeons Pakistan-accredited teaching hospitals in Karachi. Items were grouped a priori into Knowledge (4 items), Attitude (8 items), and Practice (9 items) domains. Composite scores were calculated (1–7 scale), and internal consistency was assessed with Cronbach’s alpha. Between-group differences were evaluated using Mann–Whitney U or Kruskal–Wallis tests. An ordinal logistic regression was used to identify predictors of Practice category, and a linear regression sensitivity analysis was performed using continuous Practice scores.

Results

Eighty-two respondents participated (51% public, 49% private institutions). Internal consistency was acceptable for all domains (α = 0.73–0.8). Mean (SD) scores were: Knowledge 6.10 (0.77), Attitude 5.99 (0.73), and Practice 5.24 (0.87). Over 90% agreed with actions such as measuring departmental carbon footprints and favoring well-maintained circle systems. No significant differences in scores were observed by institution type or designation categories. In regression analysis, Attitude was a significant predictor of Practice. The modest sample size may limit statistical power and the stability of regression estimates.

Conclusion

Anaesthesiologists in teaching hospitals in Karachi demonstrated high knowledge and positive attitudes toward sustainable anaesthesia, but reported practice was lower and may be overestimated by self-report. Given the modest sample size and single-city design, caution is warranted in generalizing these findings beyond similar urban academic settings.

Trial registration

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12871-026-03842-8.

Background

Healthcare is responsible for 4–5% of global greenhouse-gas emissions [1]. Within the health sector, perioperative care is a modifiable source. Inhalational anaesthetic agents and gases (e.g., desflurane, isoflurane, nitrous oxide) have a high global-warming potential and are vented unchanged into the atmosphere [2]. Reducing their use from perioperative care thus offers both local and global benefits.

Anaesthetic agents also vary in their global-warming potential. Nitrous oxide is being phased out in many high-income countries, yet remains widely used in Pakistan as well as in other Lower Middle Income Countries (LMICs) because of its low cost and ready availability. In a tertiary-centre audit from Pakistan, nitrous oxide cost was 5% but accounted for approximately 75% of operating room anaesthetic emissions [2]. The operating rooms also contribute disproportionately to hospital waste through energy-intensive single-use devices and improper waste segregation [3].

These emissions can be reduced by conservative use of agents, using low-flow anaesthesia and end-tidal control, without compromising patient safety. The World Federation of Societies of Anesthesiologists (WFSA) sets out global principles for environmentally sustainable anaesthesia [4], and other professional colleges (e.g., ANZCA PS64) provide operational recommendations that departments can adapt to local constraints [5]. However, translating guidance into routine practice remains challenging in lower-middle-income settings, where competing clinical priorities, procurement limitations and infrastructural constraints can hinder uptake [1].

Data on clinicians’ knowledge, attitudes and behaviours regarding sustainability in LMIC perioperative systems are limited, and in Pakistan it remains unclear whether anaesthesia professionals working in teaching hospitals perceive global sustainability principles as relevant and actionable in their local context. Our objective was to use the WFSA recommendations as a benchmark to quantify anaesthesiologists’ knowledge, attitudes, and practices regarding sustainable anaesthesia. Our secondary objective was to explore subgroup differences (institution type, designation) and to test whether attitude and knowledge can independently predict practice. For the latter, we hypothesized that a knowledge–practice gap exists and explored whether attitude and knowledge were independently associated with practice.

Methods

We conducted a cross-sectional survey of anesthesiologists in College of Physicians and Surgeons Pakistan (CPSP)-accredited teaching institutions in Karachi. Exemption from review was granted by the Institutional Ethics Review Committee (ERC # 2023-9241-27283). Data were collected between April 2024 and May 2025. Participation was voluntary and anonymous. Eligible participants included consultants, trainees, and staff anaesthesia medical officers working in teaching hospitals in Karachi.

A purposive, multi-channel sampling approach was used. Heads of department at eligible institutions were initially contacted by email and invited to circulate the survey link to relevant anaesthesia staff within their departments. Where no response was received by email, follow-up contact was made via WhatsApp. If a head of department remained unreachable or did not respond, focal authors known to the study team were engaged to support survey circulation within their institutions by email. Of the 16 invited institutions, 11 participated. Based on institutional staffing data, the eligible anaesthesia workforce across participating centres was estimated at approximately 250 individuals. The survey tool was adapted from the World Federation of Societies of Anesthesiologists (WFSA) Global Working Group consensus statements on environmentally sustainable anaesthesia [4]. It comprised three sections. An introduction that provided a brief explanation of the study and electronic consent to participate. The second section provided information on age, gender, job designation, institution, and years of anaesthesia practice. The third section was on agreement with the WFSA Global Working Group’s consensus statements. Items were grouped a priori into three domains: Knowledge (4 items), Attitude (8 items), and Practice (9 items). Each item used a 7‑point Likert scale (1 = Strongly disagree … 7 = Strongly agree).

Statistical analysis

A total of 153 survey responses were received. Ninety-six responses were complete. Fourteen complete responses were excluded because they were from outside Karachi or from non-teaching hospitals, leaving 82 responses in the final analytic dataset. Because the survey was distributed through multiple digital channels, the exact number of individuals who received the survey link could not be determined. Based on institutional staffing data, the eligible anaesthesia workforce was estimated at approximately 250 individuals, corresponding to an estimated response proportion of 32.8%. A de-duplication audit was done of the final dataset by screening for exact matches across all recorded variables; no duplicate records were identified. The final analytic dataset contained complete data for all variables included in the analysis. For each respondent, we computed mean domain scores on the 1–7 scale, Knowledge mean, Attitude mean, and Practice mean, with higher values indicating greater agreement with the construct. We also calculated an overall Total mean across all items. Demographic variables were age, total years of anaesthesia experience, institution type (Public/Private), and designation (Consultant/Trainee/Staff).

Internal consistency for each domain was assessed using Cronbach’s alpha (α). To further examine scale performance, we also reviewed item–total correlations and alpha-if-item-deleted statistics for each domain. Because the domain scores were derived from Likert-scale responses and showed non-normal distributions on visual inspection, we used non-parametric tests: Mann–Whitney U for two-group comparisons (public vs. private) and Kruskal–Wallis for comparisons across designation categories. Two-sided p-values are reported with α = 0.05.

To respect the ordered nature of the outcome, Practice was categorized into Low (< 4.5), Medium (4.5–5.5), and High (> 5.5). These thresholds were specified a priori to aid interpretability around the mid-range of the 1–7 scale and to support ordinal regression modelling. We then fit a proportional-odds ordinal logistic regression with predictors Attitude mean, Knowledge mean, age, total experience, institution type (Private vs. Public), and designation (Trainee vs. other). We report odds ratios (ORs) with 95% confidence intervals (CIs), p values, and model fit indices (log-likelihood, AIC); the proportional-odds (parallel-lines) assumption was checked. As a sensitivity analysis, we also modelled Practice as a continuous outcome using multiple linear regression with the same predictors to assess consistency of findings across modelling approaches. We also examined correlations among key continuous variables and assessed multicollinearity in regression models using variance inflation factors (VIFs).

To quantify the mismatch between knowing and doing, we defined KP Gap = Knowledge mean − Practice mean. We fitted a multiple linear regression with HC3 robust standard errors using the same covariates. We report coefficients (β), 95% CIs, p values, and R²; residual and heteroscedasticity checks did not indicate major violations.

Results

A total of 153 survey responses were received. Of these, 96 were complete. Fourteen complete responses were excluded because they originated from outside Karachi or from non-teaching hospitals. The final analytic sample comprised 82 respondents from 11 teaching institutions; five invited institutions did not contribute any included responses. Because the survey was distributed through multiple channels, the exact number of individuals reached could not be determined. Based on institutional staffing data from the 11 participating centres, the eligible pool was estimated at 250 anaesthesiologists, corresponding to an estimated response proportion of 32.8%.

Internal consistency was acceptable to good (α Knowledge = 0.792, α Attitude = 0.808, α Practice = 0.735). For the four-item Knowledge domain, item–total correlations ranged from 0.54 to 0.67, supporting acceptable coherence among items despite the limited number of items. Item–total correlations ranged from 0.33 to 0.68 for the Attitude domain and from 0.22 to 0.65 for the Practice domain. Domain means (1–7) were Knowledge 6.10 (SD 0.77), Attitude 5.99 (0.73), Practice 5.24 (0.87); the overall total mean across items was 5.76 (0.68). Distributions are shown in Fig. 1, and summary statistics and reliability appear in Table 1.

Fig. 1.

Distribution of domain scores. Legend: Boxplots of Knowledge, Attitude, and Practice domain means (1–7 scale; n = 82). Central line = median; box = IQR; whiskers = 1.5×IQR; points = outliers. This figure illustrates the core pattern, Knowledge and Attitude are high, while Practice is lower on average

Table 1.

Reliability and domain scores (1–7 Likert scale)

Domain	Items (n)	Cronbach’s α	Item–total correlation range	Mean (SD)
Knowledge	4	0.792	0.54–0.67	6.10 (0.77)
Attitude	8	0.808	0.33–0.68	5.99 (0.73)
Practice	9	0.735	0.22–0.65	5.24 (0.87)

Item–total correlation ranges are provided as an additional indicator of internal consistency. Domains were grouped a priori on conceptual grounds; formal construct validation was not performed

No statistically significant differences were observed by institution type (public vs. private) on Knowledge (p = 0.433), Attitude (p = 0.706), Practice (p = 0.234), or Total scores (p = 0.332). Likewise, no statistically significant differences were observed between designation categories on Knowledge (p = 0.258), Attitude (p = 0.251), Practice (p = 0.320), or Total scores (p = 0.314).

Higher Attitude mean was independently associated with being in a higher practice category (OR 6.18, 95% CI 2.29–16.68, p < 0.001). Because Attitude was measured as a mean score on a 1–7 scale, a one-unit increase reflects a substantial increase in average endorsement of pro-sustainability attitudes across the attitude items. Knowledge and demographic covariates (age, experience, institution type, designation) were not statistically significant. Model fit was adequate (log‑likelihood − 71.55; AIC 159.1). The distribution of Practice categories is shown in Fig. 2; full model estimates are presented in Table 2. Knowledge and Attitude mean scores were positively correlated (r = 0.72;Supplementary Table S2). However, multicollinearity was not considered problematic, with low variance inflation factors in the linear sensitivity model (VIF 2.17 for Knowledge and 2.15 for Attitude). In a sensitivity analysis using multiple linear regression with Practice mean as a continuous outcome, Attitude remained a strong and statistically significant predictor of Practice (β = 0.80, 95% CI 0.51 to 1.10; p < 0.001). Knowledge, institution type, and designation were not significantly associated with Practice. Age showed a borderline positive association (β = 0.025, 95% CI -0.001 to 0.051; p = 0.062), while total experience was not significant (β = -0.027, 95% CI -0.058 to 0.005; p = 0.100). The model explained 45.3% of the variance in Practice scores (R² = 0.453). These findings were consistent with the ordinal regression model (Supplementary Table S1).

Fig. 2.

Distribution of practice categories. Legend: Bar chart showing counts of respondents classified as Low (< 4.5), Medium (4.5–5.5), and High (> 5.5) Practice means (n = 82). Labels show frequencies

Table 2.

Proportional-odds ordinal logistic regression predicting practice category (Low/Medium/High)

Predictor	OR	95% CI	p-value
Attitude mean	6.18	2.29–16.68	< 0.001
Knowledge mean	0.87	0.34–2.20	0.762
Age (years)	1.06	0.98–1.15	0.146
Total experience (years)	0.93	0.84–1.02	0.128
Institute = Private	1.68	0.66–4.23	0.276
Designation = Trainee	1.18	0.42–3.32	0.748

Model fit: Log-likelihood = − 71.55; AIC = 159.1. Proportional-odds assumption reasonably met (parallel-lines test not significant). Cut-points for Practice: Low < 4.5; Medium 4.5–5.5; High > 5.5. Interpretation: Higher Attitude is independently associated with being in a higher Practice category; Knowledge and demographics were not significant predictors after adjustment

The regression model explained a large portion of the variation in the Knowledge–Practice (KP) gap (R² = 0.68). Higher knowledge scores were associated with a larger observed gap between knowledge and practice (β = +1.32; 95% CI 0.67–1.97; p < 0.001), whereas more positive attitudes were associated with a smaller observed gap (β = −1.03; 95% CI − 1.56 to − 0.49; p < 0.001). These findings should be interpreted as associations rather than causal effects. Figure 3 shows how the KP gap varied across different staff designations.

Fig. 3.

Knowledge–Practice (KP) gap by designation. Legend: Boxplots of KP Gap = Knowledge mean – Practice mean across Consultant, Trainee, and Staff categories (n = 82). Dashed line at 0 indicates no gap; positive values indicate higher knowledge relative to reported practice

Discussion

In this cross-sectional survey of anesthesiologists from eleven Karachi teaching hospitals, participants reported good knowledge and attitudes toward environmentally sustainable anaesthesia but comparatively lower reported implementation in practice. These findings suggest that translating awareness into action may be more closely associated with attitudes than knowledge alone. On further analysis, higher knowledge was associated with a larger observed knowledge–practice gap, whereas more positive attitudes were associated with a smaller observed gap. Together, these findings suggest that attitudes may be more closely linked to reported uptake than knowledge alone in this sample. Although the observed odds ratio was large, the predictor was measured on a 1–7 averaged scale; therefore, a one-unit increase represents a substantial shift in overall attitude. This likely explains part of the magnitude of the association. In addition, correlation and multicollinearity checks suggested that the Attitude effect was not simply an artefact of overlap with Knowledge.

Closing this gap may require interventions that address both motivational and structural factors influencing sustainable choices. The observed gap between knowledge, attitudes, and reported practice suggests that awareness alone may be insufficient to support implementation. In this context, future work should examine which institutional and behavioural factors most strongly influence sustainable anaesthesia practices in teaching hospitals. Potential strategies described in prior literature include education linked to behaviour change, audit and feedback, locally adapted sustainability guidance, and institutional systems that support lower-emission practices [4, 6, 7].

Our results, showing high knowledge and attitudes but lower reported practice, are consistent with reports from other settings in which awareness appears to outpace behaviour change in perioperative sustainability [8, 9]. Reports from other settings suggest that local innovation may support decarbonisation while strengthening resilience. The Kyabirwa Surgical Centre integrates water–energy systems and achieves high uptake of low-emission techniques [10], and other LMIC-led initiatives in intensive care have paired stewardship with mutual learning across settings [1, 11] However, the feasibility and generalisability of such approaches in this context require further study. Broader system-level approaches, including integration of climate considerations into surgical planning, have been proposed elsewhere, but their applicability in this context requires further evaluation [12] and aligns with calls to decarbonize health systems as part of universal health coverage [13].

The strengths of our study are that it adds to South Asia–specific evidence on anaesthesia sustainability. Our scales were reliable and analysed outcomes with an ordinal model appropriate for ordered data. By reporting results at the item level, we have identified specific behaviours with low uptake, clear targets for education, policy, and procurement interventions. Our study has several limitations. First, the cross-sectional, self-reported design is particularly vulnerable to social desirability bias, as respondents may have overstated environmentally favourable knowledge, attitudes, and especially reported practices. This is important in the present study because Knowledge and Attitude scores were high, and reported Practice may reflect aspirational or socially preferred responses rather than actual routine behaviour. As a result, the true gap between favourable attitudes and actual day-to-day practice may be larger than observed in this survey. Second, the modest sample size (n = 82) limits statistical power and may reduce the stability and precision of regression estimates, as reflected in relatively wide confidence intervals. Third, the inclusion of multiple predictors in regression models with a limited sample increases the risk of overfitting; therefore, these analyses should be considered exploratory and interpreted with caution. Fourth, the study was conducted in teaching hospitals within a single metropolitan area and therefore reflects an urban academic context that may be better resourced than non-teaching, rural, or lower-resourced settings. This limits generalizability to other institutions within Pakistan and to LMIC settings more broadly. The estimated response proportion was 32.8%, and selection and non-response bias remain possible. Because the study was cross-sectional, the direction of observed associations cannot be established; reverse causality and unmeasured confounding remain possible. Although internal consistency was acceptable, the domain structure was based on a priori conceptual grouping rather than formal psychometric validation; therefore, the Knowledge, Attitude, and Practice domains should be interpreted as related but not definitively distinct constructs. Given the modest sample size, we did not perform exploratory or confirmatory factor analysis to formally assess dimensionality. Finally, the absence of objective practice measures further constrains interpretation. Nevertheless, the findings remain valuable as one of the first local datasets exploring this domain and may guide future multi-centre studies in Pakistan. Future work should triangulate self-reports with objective metrics (e.g. machine logs and procurement data) and broaden sampling across provinces. In addition, qualitative studies should delineate barriers (procurement, defaults, consumable availability) and test strategies such as departmental champions, protocolized low-flow defaults, and incentive-compatible procurement policies.

Conclusion

Our study of anaesthesia providers in teaching hospitals in Karachi found high levels of knowledge and generally positive attitudes toward environmentally sustainable anaesthesia, but comparatively limited reported adoption of such practices. These findings reflect an urban academic setting in Pakistan. The observed knowledge–practice gap suggests that institutional support and behavioural factors may be important alongside awareness, although these cross-sectional findings should not be interpreted causally.

Abbreviations

ANZCA

Australian and New Zealand College of Anaesthetists

CPSP

College of Physicians and Surgeons Pakistan

HICs

High–Income Countries

KP Gap

Knowledge–Practice Gap

LMICs

Low–and Middle–Income Countries

NSOAP

National Surgical, Obstetric, and Anaesthesia Plan

WFSA

World Federation of Societies of Anaesthesiologists

Authors’ contributions

AN contributed to institutional liaison, literature search, survey distribution coordination, data collection, preliminary data analysis, manuscript write-up support, and proofreading. SAD was involved in the supervision of data collection, data analysis planning, interpretation of results, drafting of manuscript, critical revision, and final approval. AA contributed to the conceptualisation of study, survey design, data management, data cleaning, accuracy checks, interpretation support, preparation of tables and figures, and manuscript review. FN was involved in participant coordination, distribution of survey link across institutions, data collection, and manuscript editing. AAS contributed to literature review, drafting selected sections of introduction and discussion, and proofreading. AM provided statistical support, verification of analysis, interpretation assistance, and manuscript revisions. ZR contributed to the facilitation of survey access at training centres, data contribution, and critical review. SS was involved in the validation of results interpretation, and manuscript proofreading. AJ supported in data collection, database organisation, and manuscript editing. FK provided senior supervision, methodological oversight, critical revision of manuscript, validation of final content, and final approval.

Funding

None to declare.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study adhered to the Declaration of Helsinki and was exempted from ethical review by the Ethics Review Committee of The Aga Khan University (ERC # 2023‑9241‑27283). Informed consent to participate was obtained from all of the participants in the study at the time of enrolment.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.