Views and attitudes of healthcare professionals on do-not-attempt-cardiopulmonary-resuscitation in low-and-lower-middle-income countries: a systematic review

Abstract

Background

Healthcare Professionals (HCPs) are important stakeholders and gatekeepers in resuscitation decision-making. This systematic review explored the views and attitudes of HCPs on do-not-attempt-cardiopulmonary resuscitation (DNAR) in low-and-lower-middle-income countries (LLMICs).

Methods

PubMed, EMBASE, PsycInfo, CINAHL, Cochrane library, Scopus, and Web of Science were searched from 01-Jan-1990 to 24-February-2023. Empirical peer-reviewed literature exploring views and attitudes of HCPs on DNAR for adult patients (aged ≽18 years) in LLMIC were included. No restriction on empirical study designs was imposed. Two independent reviewers performed screening, data extraction and critical appraisal. Hawker’s tool and Popay’s narrative synthesis were used for critical appraisal and data synthesis respectively. Review findings were interpreted using Cognitive Dissonance theory (CDT).

Results

Of the 5132 records identified, 44 studies encompassing 7490 HCPs were included. The median Hawker score was 28 with 27% studies having low risk of bias. Three themes emerged. 1: Meaning-Making of DNAR construct. Most HCPs agreed that DNAR avoided inappropriate resuscitations, needless suffering and allowed fair allocation of resources. However, there was a lack of consensus on DNAR timing. 2: Barriers and Facilitators. Sociocultural norms, lack of legal clarity, organisational policies, societal and family views, religious and ethical beliefs, and healthcare providers’ presuppositions often hindered DNAR practice. HCPs had inconsistent religious and ethical beliefs about DNAR. 3: Tensions and complexities of contemporary practice. HCPs expressed fears, concerns, guilt and distress while recommending DNAR. HCPs differed on involving patients. The DNAR practice was arbitrary and suboptimal like informal DNAR orders, pretended and symbolic CPRs.

Conclusion

Most HCPs in LLMICs viewed DNAR as essential However, they faced barriers to DNAR implementation at macro-(law, sociocultural norms), meso-(organization) and micro-(HCP- and family views) levels. These barriers contributed to HCPs’ fears, concerns and distress concerning DNAR. The CDT provided the lens to link HCPs cognitions, affect and behaviour into a chain of events that explained suboptimal resuscitation practices.

Trial registration

CRD42023395887.

Introduction

Cardiopulmonary resuscitation (CPR) refers to interventions performed to restore circulation and breathing during a cardiopulmonary arrest [1]. It can be life-saving, but has its limitations and adverse effects [2]. Despite the evidence that CPR may be futile and contribute to potentially avoidable suffering at end-of-life (EOL), it is often used injudiciously in patients with frailty and chronic life-limiting illnesses (CLLI) [3, 4].

Do-not-attempt-resuscitation (DNAR) should be a shared decision between the healthcare professionals (HCPs), patients and/or caregivers to not perform CPR on grounds of refusal by the patient, medical futility, risks overweighing benefits or patients’ best interests [5,6,7]. The evidence suggests that DNAR practice is suboptimal and varies across settings [8]. HCPs views and attitudes have been shown to influence resuscitation decision-making, which at times are incongruent with the patients preferences [3, 9,10,11].

Low-and-lower-middle-income-countries (LLMICs) included in this review are in accordance with the World Bank categorization and their list is provided in Supplementary-file 1 [12]. In these geographical settings, there is a significant burden of non-communicable diseases, cancer, end-stage organ impairment and health-related suffering [13]. Formal processes in DNAR practice are uncommon, delayed or associated with implementation challenges [14, 15]. CPR precedes most hospital deaths despite poor resuscitation outcomes like survival and hospital discharge [14, 16].

Knowing views and attitudes of HCPs on DNAR can help understand context-specific barriers and facilitators to DNAR practice. A preliminary scoping search did not identify any existing or ongoing systematic review (SR) exploring views and attitudes of LLMIC’s HCPs on DNAR. Published synthesised literature lacks contemporality and is skewed towards high-income countries (HICs), necessitating the conduct of this review [2, 8, 17,18,19,20,21,22,23,24,25,26].

Methods

The review question was: What are the views and attitudes of HCPs on DNAR in LLMICs? The PICo framework (HCPs – Population, views and attitudes on DNAR—phenomenon of Interest, and LLMICs—Context), was used to develop the review question. FINER (Feasible, Interesting, Novel, Ethical and Relevant) criteria, as recommended by the Cochrane handbook of systematic reviews, were referred to while formulating the review question [27]. The preliminary scoping work validated the feasibility of the review by confirming the availability of evidence addressing the review question and its novelty. The research priority setting meeting with academic and local supervisors confirmed that the review topic wasinteresting, ethical and relevant. This SR followed the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) standards [28]. The protocol was registered with the PROSPERO (CRD42023395887).

Eligibility criteria

The inclusion and exclusion of studies in this review were according to the eligibility criteria provided in Table 1.

Table 1 Inclusion and exclusion criteria

Information sources

Four subject-specific [PubMed, CINAHL (EBSCOhost), Embase (OvidSP), and PsycINFO (OvidSP)], and three multi-disciplinary databases [Scopus (Elsevier), Web of Science (Clarivate), and Cochrane Library] were searched. Peer-reviewed empirical literature published in English was considered for inclusion. The search was limited to publications from 01.01.1990 to 24.02.2023 to reflect the current views and attitudes on DNAR. This timeframe was chosen as much of the global perspectives on DNAR emerged after the passage of the US Patient-Self-determination Act in the United States of America in 1990 [19, 31]. Bibliographic screening of included studies and relevant evidence synthesis and citation tracking of included studies using Google Scholar and Scopus were done to identify any additional article; which continued until no new relevant article was identified [32].

Search strategy

A three-phase search strategy was adopted [33]. A preliminary search of PubMed was performed using free text terms for the key concepts known to the researchers. The search strategy was developed iteratively, examining the article title, abstract, keywords, and thesaurus terms used to index relevant articles and exploring the search strategy used in similar reviews [8, 18, 22]. The Boolean operators OR and AND were used to combine the database-specific thesaurus and free-text words for similar and different concepts, respectively. The scoping search identified five index papers that helped test the sensitivity of the search [9, 34,35,36,37]. The search strategies were developed in consultation with a specialist health librarian and reported in detail in Supplementary-file 2 [28].

Selection process

All identified records were transported to EndNote (V.20, Clarivate Analytics, Philadelphia Pennsylvania, USA) and duplicate entries were removed. Deduplicated references were exported to Rayyan (https://www.rayyan.ai/) for screening by two independent reviewers. The reviewers first screened the title and abstracts followed by the full text reports to identify records which satisfied the predefined eligibility criteria.

Critical appraisal

Hawker’s tool was used to assess the methodological rigor of the relevant studies [38]. Each study was assessed on nine criteria, each graded on a scale from 1 (very poor) to 4 (good). The overall score ranged from 9 (very poor) to 36 (good). Hawker’s tool was chosen because of its utility in appraising mixed typology of studies in previously published palliative care SRs, structured format, brevity and availability of scoring guidance [24, 39,40,41]. Piloting was first done on ten studies by two independent reviewers to ensure consistent application [27]. The studies scoring < 20 were excluded as adopted by a previous SR [36]. Studies scoring between 30–36 were considered to have a low risk of bias [24, 42].

Data extraction

A customized Microsoft excel data extraction form was developed in consultation with all the reviewers. It had eight sections: bibliographic details, concept and context, methodology, sampling, data collection and analysis, participant characteristics, results and critical appraisal (Supplementary-file 3). Two independent reviewers (MG and UJ) piloted the data extraction form on five studies with different study designs to ensure that it captured all the relevant information [27]. Screening, data extraction and critical appraisal were done by two independent reviewers (MG and UJ) and discrepancies settled with consensus and arbitrated by a third reviewer (SRR) if required [27, 43].

Data synthesis

Popay’s narrative synthesis was used for data synthesis [44]. A preliminary synthesis was developed by tabulating textual summaries of the included studies. This synthesis helped to familiarize and contextualize the study findings [44]. Relationships within and across the included studies were explored and conceptually similar findings were grouped using mind map software (Supplementary-file 4). It was followed by inductive generation of sub-themes and themes. Heterogeneity in terms of population, context and methodology were explored. As Popay’s narrative synthesis allows using a theoretical framework [44], we used cognitive dissonance theory (CDT) to interpret the review findings [45]. The CDT has been applied previously in healthcare and in the context of DNAR [46,47,48]. The robustness of the synthesis was ascertained by inclusion of studies that met a predetermined methodological rigor (Hawker’s score > 20). Furthermore, the review findings were critically reflected upon for their limitations, implications and possible sources of bias in the studies informing the synthesis.

Results

Study selection

Database search identified 3535 records after removing 1597 duplicates. Seventy-five reports were included for full text screening. The interrater reliability (Cohen’s kappa) among the reviewers at the end of title and abstract screening was 0.74 (95% CI 0.72–0.76) and 0.81 (95% CI 0.73–0.89) respectively. We excluded 43 reports (Fig. 1and Supplementary-file 5) [49]. Another 16 reports were included through reference list and citation searching. Four of the 48 reports were linked to their primary studies [50,51,52,53,54,55,56], (Supplementary-file 6). Finally, 44 studies (48 reports) were included in this SR [9, 34,35,36,37, 50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92].

Fig. 1

PRISMA Flow Diagram

Study characteristics

The 44 studies had 7490 HCPs (Fig. 2a) involved in DNAR process (Table 2). The studies recruited participants from 16 (19.5%) out of 82 LLMICs, with 20 (45%) being published from Iran (Fig. 2b). Most studies (> 90%) were published after 2010 (Fig. 2c). Twenty-two studies had DNAR as the primary focus. The rest explored DNAR while assessing HCP’s views and attitudes on other related phenomenon (Fig. 2d). The studies were heterogenous with respect to the study design (Fig. 2e).

Fig. 2

Study classification based on the study population, country and year of publication, primary concept and study design

Table 2 Characteristics of included studies

Critical appraisal

The median Hawker score was 28 (Range 20–36). Qualitative studies (Median = 31, Range 26–36) scored higher compared to the quantitative studies (Median = 27, Range 20–33). Approximately 64% of qualitative studies compared to 10% of quantitative studies were classified as having low risk of bias (Table 2).

Review themes

Three themes and nineteen sub-themes were generated (Fig. 3). Given the complexity and magnitude of findings, Table 3 demonstrate examples of how key study findings translated into elementary themes and Table 4 depicts study-wise contribution to different sub-themes respectively. As is evident from Table 4, multiple studies, both qualitative and quantitative contributed to each sub-theme, precluding the possibility that exclusion of 1–2 studies to significantly alter the synthesis findings.

Fig. 3

Thematic Map

Table 3 Review themes, sub-themes and explanatory narratives

Table 4 Study-wise contribution to themes and sub-themes [9, 34,35,36,37, 50, 53, 55, 57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92]

Theme 1: meaning-making of DNAR construct

Meaning-making refers to how individuals interpret and comprehend the world and the self and its relation to cognition and behaviour [93]. This theme describes HCPs’ views on the understanding of the DNAR process, timing, implications, benefits and burden, and its perceived role in mitigating suffering and resource allocation.

Awareness and understanding

Most physicians knew the term DNAR [35, 37, 83, 84]. Awareness was significantly associated with intensive care unit (ICU) training [37]. However, HCP’s understanding of DNAR varied across country settings [74, 83, 88]. A few perceived DNAR as euthanasia [9], passive euthanasia [75, 78, 84], killing a person [36], withholding life sustaining treatments (WLST) [37] and end-of-life (EOL) [69]. DNAR knowledge improved with education [69, 74]. While work experience was reported as the major contributor to awareness, only 11.2%−27.1% of HCPs reported formal education as their source of information [83, 88]. Although education enabled younger HCPs to be more aware [35, 69], it did not translate into practice in a hierarchical workplace culture [69].

Perceived implications

HCPs perceived DNAR as only withholding CPR [69, 88, 90] and not withholding other life-sustaining treatments (LSTs) [88, 90], do-not-treat [88] or no care [61, 88, 90]. Nurses believed that patients initiated on DNAR should receive either same or higher quality of care [61, 75, 79, 88, 90]. HCPs, thus, preferred to continue vital signs monitoring, symptom management, emotional support, bleeding control, personal hygiene, pressure ulcer prevention, oxygen, chest physiotherapy, feeding, IV fluids, ventilatory support, inotropes, antibiotics, blood products, diagnostic imaging and dialysis [61, 69, 75, 79, 86, 88, 90]. Palliative treatments, personal hygiene and pressure ulcer prevention were the most commonly offered treatments and diagnostic tests, and tracheal intubation were often withheld [75]. Nurses were often unclear about the degree of appropriate care and physicians were more likely to consider feeding, antibiotics, and dialysis as appropriate supportive measures [86, 88]. HCPs’ understanding differed when distinguishing DNAR from the withdrawal of LSTs and was significantly influenced by ICU and EOL training [37, 88].

Timing

HCPs had mixed views on the timing of DNAR. While some believed in early decision-making [70, 87, 88], others conflated DNAR with EOL [34, 57, 66, 69, 71, 89]. This often led to delays in decision-making with patients losing their capacity to make considered decisions [68, 76]. This in turn was perceived as a barrier to DNAR discussions where HCPs had to face the dilemma of determining surrogate decision-makers [76]. Most HCPs believed that decisions in an unconscious patient require family’s consent [37, 70]. Only a few felt that HCPs or families could make these decisions unilaterally [37]. Appropriate timing of DNAR initiation often correlated with previous EOLC training [37].

Benefits and burdens

HCPs considered CPR in patients unlikely to survive as futile [53, 55, 64, 75, 79] Futility was articulated both physiologically and qualitatively and not just survival [9, 64]. HCPs felt their role should transcend fostering hope or prolonging life [55, 75] and believed that CPR should be applied selectively [58, 63, 73]. They advocated for DNAR in patients with advanced diseases, terminal cancers, permanent brain impairment, critically ill or imminently dying [34, 53, 57, 60, 61, 63, 71, 74, 75, 80, 88, 92]. They wished DNAR for themselves and their families if terminally ill [75, 79] or if CPR was inappropriate [34, 57, 71]. HCPs believed that DNAR helped clarify treatment goals at EOL [34, 55, 57, 69, 71, 75, 79]. Malignancy, advanced age, multiple organ dysfunction and respiratory failure were the most common considerations informing HCP’s decision [74, 90, 92]. While some HCPs considered CPR to be of limited value in elderly [53, 74, 75, 92], others (44%−56.1%) disagreed with this concept of ageism [55, 79, 90].

Contrastingly in some studies, HCPs disagreed with the concept of futility [55, 84]. They preferred CPR for their families even when terminally ill [55, 80] and would choose interventions with even minimal chance of improving survival at EOL [80, 84].

Mitigates suffering

HCPs viewed CPR as inconvenient for terminally-ill patients and DNAR as a means of comfort, protecting patients and their families from unnecessary suffering [34, 55, 57, 61, 63, 64, 69, 71, 75, 79, 83, 92]. HCPs anticipated that by avoiding unnecessary resuscitations, DNAR would also reduce their workload and fatigue [67, 69]. However, in one study, nurses clarified that the DNAR decision should be based upon patients’ prognosis and functional status and not just upon the patient’s discomfort [90]. While one study showed [80] Iranian nurses held negative attitudes toward DNAR as a means of avoiding suffering and preserving dignity, eight studies of HCPs revealed a positive attitude [34, 57, 61, 64, 71, 75, 79, 92].

Allows fair allocation of resources

HCPs believed that futile CPR had adverse financial impact on the family and country [9], while DNAR had cost-and resource-saving benefits [9, 67, 69] They found the economic burden of keeping terminally ill patients alive unjustifiable [55, 75, 79]. Iranian families cited limited resources as the most common reason for accepting DNAR [64, 83]. One study showed that family’s socio-economic status influence HCPs’ decision-making [66], while others did not [90, 92]. HCPs in some studies viewed DNAR as a medical decision guided by patient preferences, unaffected by healthcare costs, socioeconomic status, critical care bed availability or duration of hospital stay [63, 88, 90].

Theme 2: barriers and facilitators

The terms barriers and facilitators were used to describe the interplay of various generative mechanisms and the interdependence of factors within social systems that facilitated or hindered DNAR implementation in LLMICs [94].

Individual factors

Most HCPs felt that DNAR was essential and a right of terminally ill patients [34, 35, 50, 57, 58, 62, 63, 67, 69, 71, 75, 77, 78, 83, 85, 87, 88]. Physician's acceptance varied from 48% (Iran) to 100% (Pakistan, Philippines) [63, 82]. In contrast, in a few studies, HCPs had a negative view on DNAR [35, 37, 72, 80, 83]. Nurses were less likely than physicians to consider DNAR as a right of terminally ill patients [50] or consider family’s consent as necessary [70]. HCPs from LLMICs were less likely to accept DNAR than those from HICs [73, 81, 82]. HCPs in some studies considered DNAR to be either non-applicable, not needed, of limited relevance or underused in their countries [69, 72, 91].

HCPs had better acceptance of DNAR compared to patients and families [71]. In a few studies, better DNAR acceptance was associated with higher education [34, 37, 71, 74], male gender [37, 78], work experience [78, 85], and DNAR knowledge [74]. Most studies however did not mirror these associations [34, 35, 37, 74, 80, 81, 85, 88, 90]. Longer professional experience [57, 69], working in oncology [50] and higher working hours [78] correlated with positive attitude towards DNAR. History of COVID-19 infection or death of a relative with COVID-19 among HCPs was associated with a favourable attitude towards DNAR in COVID-19 patients [78]. The trainees were apprehensive to initiate DNAR orders and abdicated decision-making responsibility to senior doctors [61, 89]. Lack of education in communication and EOLC [69] and prognostication [89] influenced DNAR discussions, with only 21.8% doctors expressing confidence in discussing code status independently [58]. HCPs’ age, ICU type, marital and parental status, courses in CPR, ethics or critical care, experience of caring for patients with or implementing DNAR orders were not associated with their attitude towards DNAR [34, 74, 80, 81, 85, 90].

Family views

Review findings indicate that HCPs considered family views and socioeconomic status as both barriers and facilitators to DNAR decision-making [9, 64, 69, 72]. Family requests for non-disclosure were often acceded by the HCPs as they were often the first point of contact [66, 69, 70, 89] and primary caregivers at EOL [69]. Moreover, preferences to discuss with patients, family, or both were influenced by HCPs' country of origin and training [66, 89]. One study found that Egyptian nurses felt that family preferences influenced their DNAR practice [90].

HCPs (trainees more than physicians) reported families as the most significant barrier to DNAR [9, 64, 89]. Families in Iran and Pakistan were closely knit, emotional and fearful of death [9, 72]. The families not informed of prognosis were more insistent on CPR [9, 69, 72] DNAR was viewed as euthanasia and giving up on loved ones, which families felt guilty to approve [9, 69, 72]. Family disapproval, education level, language barrier, intra-family conflicts, and doctor in the family impeded DNAR discussions [89]. Notably, they considered both the extremes of families abdicating decision-making on them as well as deciding against their recommendations as hinderances to DNAR decision-making, which indirectly reflected their predilection for shared decision-making [89].

Affluent families sometimes demanded unhelpful interventions, which HCPs viewed as a self-serving behaviour [64, 72]. Contrastingly, poor families due to inadequate resources often acceded to DNAR decisions [9, 64]. Families agreed for DNAR to preserve patient’s dignity [64, 72] which was the most common reason among Iranian families [64].

Religious beliefs

HCPs’ ambivalent religious beliefs influenced their views on DNAR [55, 69, 75, 79]. Some viewed death as a predetermined fate [64], inappropriate CPRs as 'Sin' [9, 64] and DNAR not in contradiction with their religious beliefs [9, 34, 57, 63, 64, 71]. However, HCPs who believed in miracles, God-Centredness of life and death, suffering as a means to absolve sins and fear of committing sins were apprehensive of documenting DNAR [9, 36, 37, 55, 59, 64, 69, 72, 79, 80, 83]. Most studies reporting religious beliefs incompatible with DNAR were from countries where majority practiced Islam. The Iranian HCPs had conflicting religious beliefs between the inevitability of death (pro-DNAR views) and belief in miracles (anti-DNAR views) [9, 64]. HCPs from Iran [9] and Pakistan [72] described strong anti-euthanasia views while expressing their views on DNAR, with one explicitly equating DNAR with euthanasia in Pakistan [72]. No difference in attitude was observed between the Shia and Sunni nurses [79]. One study showed religiosity impacting HCPs attitude with orthodox more likely than secular to forego CPR [81].

Society and socio-cultural norms

Iranian and Palestinian HCPs considered DNAR as culturally unacceptable [34, 55, 60, 75, 79, 80]. In Gibbs et al., a physician described DNAR as culturally prohibited in Uganda [72]. Geographical variation was evident as contrary to countries like Iran and Palestine with strong sanctity of life beliefs, physicians from Nepal reported DNAR not in conflict with their cultural beliefs [57]. Doctors’ role was identified as life-saviours [69] and “symbol of hope and not death” [89]. Ozer et al. reported that physicians from Indonesia emphasized less on the quality of life (QoL) compared to physicians from Mexico, while three studies from Sri Lanka [69], Lebanon [66] and Palestine [55] reported that most physicians and nurses considered QoL more important that quantity of life.

Organisational challenges

HCPs reported lack of formal protocols, systems, national guidelines, DNAR policies, standardized forms, and institutional support for DNAR implementation [9, 35, 53, 64, 65, 69, 89, 90]. Physicians cited time constraints as a barrier to DNAR discussions [66, 69, 89]. Pakistani physicians (trainees more than physicians) reported lack of administrative support, trained nurses and hospital policies as barriers [89]. Physicians preferred counselling room for DNAR discussions with lack of designated space being a barrier [89]. Others reported hierarchical [69], stereotyped [9, 64] and private healthcare system [72] in LLMICs as barriers to DNAR implementation. While Egyptian nurses disagreed that hospital policy influenced their decisions [90], Iranian nurses reported being influenced by the hospital [75, 79] and peers [55, 75, 79] for implementing DNAR orders.

Ethical and moral conundrums

HCPs had contrasting moral views on DNAR. Some viewed futile CPR as a violation of autonomy, beneficence and nonmaleficence, harming both body and soul [60, 64]. Non-maleficence was the most important moral principle informing HCPs views [64]. DNAR was considered as ethical and congruent to human dignity and autonomy [9, 34, 57, 63, 71, 83]. Iranian HCPs described inability to implement DNAR as morally challenging [60]. Contrastingly, in few studies, HCPs were unable to ascertain the morality of DNAR [61] and some considered it immoral [36, 80, 83] and inhumane [37].

Legal status

Lack of legal status was the most consistent and important barrier to DNAR implementation [9, 53, 55, 59, 61, 64, 65, 67,68,69, 72, 75, 79, 90, 91]. In one study, ED physicians reported that DNAR was not legally binding and had limited relevance in Pakistan [91]. HCPs expressed the need for a national DNAR policy, DNAR to be legalized and wished for better understanding of PC, advance directives (AD), patients’ rights and legal ramifications of DNAR [55, 64, 65, 69, 72, 75, 79]. Emergency medicine (EM) personnel felt that legalisation and policy would accord them legal protection, reduce their burden, and ensure rational use of resources [67]. Religion impacted Palestinian HCP’s attitude towards legalization of DNAR orders, with 65.3% of Muslims but none of the Christians in favour of it [55].

Theme 3: tensions and complexities of contemporary practices

A complex system has internal and external forces that can compete with or complement each other, creating tensions that may either favour or hinder the phenomenon [95]. Emotions evoked during the DNAR process, arbitrary and informal practice, threatened patient autonomy, hierarchical and non-consensual decision-making and symbolic and token CPRs are the forces that are creating tensions and threatening DNAR implementation in contemporary practice.

Emotions evoked

HCPs ubiquitously expressed fear of legal prosecution [9, 55, 59, 61, 64, 65, 68, 75, 76, 79, 89, 90, 93]. Other fears and concerns included fear of God [36], misuse potential for organ harvesting or secondary gains [83, 89], DNAR being incorrectly labelled [57, 69, 89], lack of guidance [57], patients being neglected, abandoned, treated differently, or receiving poor care after DNAR [36, 69, 89], therapeutic nihilism [69], misinterpreted as abandoning patients [69, 89] or neglect due to lack of communication skills [69], loss of public trust in the healthcare system [64], and fear of family reaction [89]. Trainees perceived them more as barriers [57, 89] and found it uncomfortable and difficult to discuss about death and DNAR [55, 75, 79, 89].

For a few, DNAR instilled feelings of despair, discomfort, depression, frustration, powerlessness, confusion, anxiety, anger and guilt [36, 57, 66, 88,89,90]. Iranian nurses described their DNAR experience as challenging with most citing religious beliefs, fear of legal prosecution, lack of formal DNAR policy and informal DNAR orders as the reasons [75]. HCPs coped with this distress by ensuring that the patients were comfortable, looked presentable, were not left to die alone and doing extra for the families [88, 90]. However, others adopted passive strategies and avoidance behaviours like anticipating improvement in patient’s condition, requesting change in assignment, and avoiding families and patients [90]. HCPs expressed the inability to practice DNAR in Iran as distressing [9, 64]. Performing inappropriate CPRs affected the quality of service delivered and led to staff burnout, depersonalisation, depression, moral distress, and guilt [9, 64].

Arbitrary and variable practice

DNAR practices were often variable with decisions made arbitrarily [9, 61, 69]. While some Iranian HCPs performed CPR routinely for all patients [9, 60, 64, 67], others reported experience in implementing or caring for patients with DNAR [34, 37, 61, 75, 80]. Educational level and DNAR knowledge were favourably associated with the DNAR decision-making [74]. There were within and between-country variations in DNAR practice. While some Indian physicians reported methods for DNAR in place [72], others felt it was not applicable in the Indian context and rarely discussed with patients [73, 87]. Some Egyptian HCPs reported practicing DNAR [35, 63] while others reported lack of written DNAR policy [90]. In Uganda, some agreed while others disagreed about having systems for DNAR [72]. Other studies reported that 20% (Sri Lanka) [37], 42.5%−56.2% (Nepal) [57] and 80% (Lebanon) [66] of HCPs had experience in DNAR decision-making. The practice of discharging terminally ill patients at EOL was common, culturally accepted and perceived as implicit DNAR in Sri Lanka [69].

Informal and verbal orders

Most HCPs believed that DNAR should be written and reviewed daily. They considered oral order illegal, with CPR being compulsory if DNAR was not documented [57, 72, 88] Contrastingly, most Egyptian nurses accepted verbal DNAR and reported poor documentation of reasons underpinning DNAR decisions, decision makers and participants in the medical records [90]. Lack of legal status and fear of legal prosecution refrained most HCPs from documenting DNAR with most orders being ambivalent, verbal or informal like placing a dot or sign on the medical records [36, 61, 64, 65, 69, 72, 74, 75, 90]. Lack of clear and written DNAR orders led to nurses not following them [55, 65] or applying wrong resuscitation measures [36] and reduced the team’s motivation to perform CPR leading to ineffective resuscitation attempts [61]. While most HCPs complied with written DNAR orders [57, 63], only 17% applied verbal DNAR orders [63]. Only few reported using written DNAR orders [72].

Threatened patient autonomy

HCPs had mixed views towards involving patients and families. Some HCPs agreed that patients and families were key stakeholders and their preferences should be considered [9, 55, 66, 70, 75, 79, 88]. Physicians believed in shared decision-making and felt that informing patients and families reduces uncertainty, allows them to come to terms with their loved-ones death, gives time to prepare and complete unfinished business [69]. Non-disclosure led to a lack of comprehension of the clinical situation contributing to distress, anxiety and non-compliance with treatments [69]. HCPs felt that communication to establish resuscitation preferences were important for patient autonomy [69].

However, other HCPs perceived themselves to be better placed to make these decisions [55, 66]. They refused to involve patients based on their preconceptions that patients did not want to be informed; informing might upset them; lead to psychological disturbances; worsen their agony, health, mood and immunity; deprive them of their will to live; or default treatments [66, 69]. Patients feared to ask questions from busy doctors who were perceived to have a ‘God-like-status’ [66, 69]. HCPs rationalised that patients lack education and are often unaware of their diagnosis, prognosis, reasons for hospitalisation, terminal phase or DNAR documentation [69]. Nurses expressed that physicians act unilaterally and undermine patients’ autonomy [60, 75].

Hierarchical and non-consensus decision making

Nurses had ambivalent attitudes towards their involvement in the decision-making process. Nurses shared that physicians tend to undermine their role by not involving them in decision-making, ignoring their views and considering them as mere decision-implementors [36, 75]. Contrastingly, nurses in some studies preferred to abdicate DNAR responsibilities to the physician [55, 70, 75, 79, 88, 90]. Iranian nurses felt it difficult to talk about DNAR [75, 79]. While nurses from Nepal agreed [88], Iranian nurses disagreed that nurses could recommend DNAR order [70]. Nurses described both non-involvement in decision-making as well as caring for patients with DNAR as painful [36]. Some nurses and physicians believed that nurses must implement DNAR orders even if contrary to nurses’, patients or families’ wishes [70]. Whereas, others opted to report their disagreement to the administration/patients’ family or decided not to follow DNAR [75].

Symbolic and tokenism

HCPs reported suboptimal practices of futile CPR being performed only for show and job safety [9, 53]. File forgery, that is CPR being recorded but not being performed [9, 74, 92], fake, symbolic, ineffective and pretended CPRs [67,68,69, 90, 92], slow code [9, 64], and intentional rib fractures [92] were practiced to avoid legal consequences. HCPs performed tokenistic CPR to comfort and satisfy family preferences [61, 64, 69, 92].

Discussion

DNAR decision-making is complex and is often influenced by HCP’s views and beliefs [20, 22, 25, 26]. Consistent with the findings of previous SRs [2, 24, 25], most HCPs in LLMICs viewed DNAR as essential. However, they faced barriers to DNAR implementation at macro-(law, sociocultural norms), meso-(organization) and micro-(HCP- and family views) levels.

HCPs, in this review, had contrasting views concerning religious beliefs, ethical dilemmas, futility of CPR and timing of DNAR consideration. Previous SRs have described similar challenges faced by HCPs [22, 25, 26]. In line with the previous research [22, 25, 26], HCPs in this review were often caught in the ethical dilemma between respecting autonomy and inflicting harm by informing about DNAR. Although many patients wanted to be involved in the decision-making [96] physicians underestimated their wish for involvement [22]. This underestimation of patient’s wish for involvement in decision-making were reported even in HICs [22]. However, there is a shift in the developed countries with landmark legal judgements and position statements reinforcing patients’ involvement in the decision-making [26, 95,96,97,98].

Our review findings showed that HCPs had conflicting views on the futility of CPR at EOL. However, across studies the descriptions of futility were subjective and ambiguous and demonstrated a lack of objective criteria to establish futility. Previous studies had also highlighted the vague and non-specific definitions of medical futility and lack of international consensus [22, 99, 100]. These non-specific descriptions make it difficult for HCPs to apply the concept of futility in the clinical practice [99] and their concern that it might be difficult to defend it in the court of law [101]. This could explain the negative attitude of some HCPs in the review who preferred to perform CPR even when deemed futile. In the absence of an international consensus, a pragmatic approach would be to use validated clinical prediction scores to estimate chances of survival with good neurological outcome which would allow the team (HCPs, patient and family) to make calculated decisions based upon the estimated success rate, other prognostic information, patient’s self-perceived QoL and overall goals of care (GOC). A consensus-based approach devoid of euphemisms put into practice by HCPs who are compassionate and trained in communication skills could reduce the conflicts and concerns (incorrectly applied, therapeutic nihilism, loss of public trust, family reaction, lack of justification and misinterpretations as abandoning) identified in the review.

The findings of this review highlight lack of legal status as the most consistent and important barrier to DNAR implementation in LLMIC. Law is a powerful mediator of human behaviour [9] and as for all healthcare decisions, also dictates EOL decisions including DNAR [102]. There are significant national variations with many western countries having specific laws, policies and procedures [103] with consequent change in HCP attitude towards DNAR [104,105,106]. However, the laws are still at a nascent stage in many Asian countries [107] and as is evident from this SR were almost non-existent in the studied LLMICs. In the absence of legal sanctions, presumption falls in favour of implied CPR [69]. Previous SRs have also iterated lack of standard guidelines, law and fear of legal prosecution as factors influencing HCPs’ decision-making [2, 24, 25]. HCPs worldwide have been reported to have poor knowledge of the law governing EOL decision-making [108]; translating into ungrounded fears and inappropriate treatments at EOL [109, 110]. This implies that a change in law in itself might not be sufficient [18, 110]. HCPs need to be educated about the existing laws, their provisions, and their application into practice.

This findings of this review show that HCPs’ in LLMIC have conflicting religious beliefs towards DNAR. They were confronted with the religious dilemma between the ‘inevitability of death’ and the ‘sanctity of life'. In Iran, strong sanctity and God-centredness of life beliefs made some HCPs to reject DNAR. However, when applied correctly, these religious beliefs prohibits one to end life and does not translate into undue and artificial prolongation of life, which was equally prohibited in Islam [20, 111, 112]. HCP’s contradictory religious beliefs in this review accord with those of Saeed et al. who in a survey showed only 29% of Muslim physicians across countries agree on the clarity of Islamic teachings on DNAR [113]. Religious doctrines like Fatwas could help clarify the distinction and have been shown to facilitate formulation and acceptance of DNAR policies in other countries with predominant Islamic faith (in context: Fatwa no. 12086 in Saudi Arabia) [114].

This SR highlights that medical overoptimism was deeply entrenched in the medical and socio-cultural context. Not just in LLMICs, death is often misconstrued as a failure of both the medical profession and the society with both patients and families insisting on aggressive management [22, 109]. Consistent with our findings, other studies also highlighted that HCPs perceived families as a source of conflict [26]. Our findings concerning family as the locus of decision-making and non-disclosure resonate with studies from Asian countries with similar socio-cultural context [23, 24, 115], often ascribed to their collectivist culture [24].

Our review findings suggest that physicians tend to undermine not just patient’s autonomy but also nurses role in the decision-making process. This professional hierarchy not only demeaned nurses’ autonomy but also deprived the team of a useful resource, who at least in theory, were most acquainted with patient’s wishes [109]. This long-standing power imbalance and dissonance generated due to non-involvement might be responsible for most nurses delegating decision-making responsibility on to the physicians. Not just LLMIC, similar abdication of DNAR decision-making responsibility had been found among the Swedish nurses which could be due to lack of clear DNAR guidelines delineating the role of different HCPs [97]. Nurses considered involvement in the decision-making and clear DNAR orders as necessary for providing good nursing care [116]. However, in resonance with our findings, others have also shown DNAR documentations to be incomplete and variable [22]. Being just implementors of the decision without being conveyed its rationality, indications, and implications precipitated distress while caring for patients with DNAR orders, as expressed by some in this review [22]. de Vries et al. (2018), while viewing resuscitation decisions through the lens of CDT, anticipated that well-documented DNAR orders (including their justifications) would reduce the dissonance discomfort associated with them [48]. Introduction of standardized forms were shown to improve the quality of DNAR documentation [18]. DNAR wristbands reduced the frequency of wrong resuscitation measures being applied, as reported by some nurses in our review [18].

Another power imbalance to emerge from the SR was the hierarchical workplace culture. In sync with our findings, lack of PC and EOLC education and training left junior doctors ill-equipped to manage these discussions, subjecting them to stress and conflicts [17]. Even if trained, their views and opinion went unheard in the hierarchical organizational structure. Education in isolation was of little benefit [22, 97] and needed to be complemented with mentoring and role modelling [117, 118].

Cognitive dissonance theory as the theoretical lens to interpret the review findings

Attitude encompasses cognitive, affective and behavioral domains that is beliefs, emotions and actions towards a person, situation or issue [119, 120]. Dissonance exists when an individuals have contrasting beliefs or actions [45]. Cognitive inconsistency was observed throughout the review findings; rationalising the use of CDT to interpret the findings. The review findings delineate HCPs' inconsistent beliefs concerning DNAR and inconsistency between the perceived utility of DNAR and actual practice underpinned by multiple contextual barriers. These inconsistencies led to dissonance in the form of psychological disturbances, moral distress and guilt among HCPs. Our findings resonate with De Vriesis claims that resuscitation decisions can create dissonance in HCPs due to incongruent beliefs [48].

Dissonance motivates an individual to either change or rationalise one of the inconsistent elements to reduce the dissonance (Fig. 4) [45]. In this review, some strategies adopted by HCPs to mitigate dissonance were changes in attitudes and denial of responsibility. They adopted suboptimal resuscitation practices like hidden DNAR code to avoid inappropriate CPR or used slow code to satisfy legal requirements and family wishes. A few changed their attitude by rejecting the concept of DNAR, while some abdicated decision-making responsibility to others. Similar suboptimal practices like informal DNAR orders were reported from some HICs, which were attributed mainly to the HCP’s lack of awareness of the guidelines [109]. Our findings also validate de Vriesis’ hypothesis that families perceive DNAR as giving up on their loved one [48]. Consequent dissonance propels them to persuade HCPs to offer unhelpful treatments leading to HCPs to perform CPRs merely to comfort the families. Evidence suggest that HCPs perform futile CPRs to accommodate families’ wishes, allowing them to come to terms, avoid confrontation and show that something has been done [25]. Timely discussion of goals of care and EOLC preferences provides families with enough justification in terms of what the patient would have wanted, thereby reducing the dissonance to do something (CPR) in an urgent situation like cardiac arrest [48].

Fig. 4

Visual representation of cognitive dissonance theory

Strengths and limitations

The exclusion of non-English studies and grey literature might have missed some relevant studies. Eight studies with Hawker’s score < 20 were excluded [121,122,123,124,125,126,127,128]. All were quantitaive questionnaire-based studies conducted in India (n = 3) [121,122,123], Iran (n = 2) [124, 125], Pakistan (n = 1) [126], Philippines (n = 1) [127] and Bangladesh (n = 1) [128]. None of them reported sample size calculation and except for two studies [122, 123]; none of them reported ethical clearance. A post-hoc analysis revealed that excluded studies contributed to one or more of the 19 generated sub-themes without any loss of depth or richness. The review explored voices of only HCPs. To ensure complete understanding, views of other stakeholders (patients, caregivers, social workers, policy makers) must be systematically reviewed. The findings of this SR should be viewed in the light that majority of the included studies were from countries where most practice Islam as their faith. This impacts the transferability of findings to non-Muslim LLMIC; represents sufficient evidence applicable to HCPs from Iran and other countries sharing similar socio-cultural and religious background.

The strength of this SR lies in its systematic, comprehensive, transparent, robust and explicit methodology and its ability to answer the review question satisfactorily. The heterogeneity of study designs and HCPs provided a rich and in-depth exploration of HCPs’ views. This review adds to the richness of the existent literature by bringing forth the socio-cultural, theological, legal and ethical barriers to DNAR implementation in LLMICs. The CDT which provided the psychological framework to link HCP’s inconsistent cognitions, dissonance generated thereof and resuscitation-preferences adopted by HCPs in LLMIC also helped to identify policy and practice recommendations to address the issues identified like formulation of standardized and context-specific DNAR guidelines, policies and forms; enactment of clear and unambiguous laws and education and training of HCPs in end-of-life care and communication skills. However, attitudes and behaviours once formed takes time to change with small and consistent steps required to bring a cultural-shift [129].

Conclusion

The review findings suggest that the majority of HCPs in LLMICs viewed DNAR as essential and necessary. They considered DNAR to avoid futile CPRs and unwarranted suffering at EOL and allow fair allocation of resources. However, lack of clarity led to inconsistent religious beliefs, ethical dilemmas and sometimes misinterpretation of DNAR with EOL. The review also highlighted some uniform barriers to DNAR practice in LLMICs like lack of legal status and standardised guidelines, HCPs’ role identification as life-sustainer, families with their death-denying and non-disclosure attitude, and cultural interpretation of DNAR as doing nothing and euthanasia. These contradictory beliefs and barriers contributed to HCPs’ fears, concerns and distress concerning DNAR. The CDT provided the lens to link HCPs cognitions, affect and behaviour into a chain of events that explained suboptimal resuscitation practices dominated by stereotyped CPRs, informal and verbal DNAR orders, symbolic and pretended CPRs in LLMICs. However, the SR findings need to be interpreted in light of its limited generalizability. About 70% of included studies were from countries where majority follow Islam as the predominant faith with 45% from only Iran. This was also reflected in the uniformity of findings across studies with each theme being contributed by a number of studies. The gaps identified in this review provide impetus for research from unrepresented countries with different socio-cultural and religious contexts.