Predictors of poor outcome from aneurysmal subarachnoid hemorrhage and an exploratory analysis into the causes of delayed neurosurgical clipping at a major public hospital in the Philippines

The provision of neurosurgical care for patients with aneurysmal subarachnoid hemorrhage (SAH) is beset with particular challenges in low- to middle-income countries (LMICs) like the Philippines. In this study located in a low-resource setting, we identify the factors that contribute to unfavorable outcomes of dependency and death. The authors retrospectively reviewed 106 patients who underwent surgery for aneurysmal subarachnoid hemorrhage in a single institution from January 2016 to September 2018. Data were obtained on exposure variables comprising patient demographics, clinical features, perioperative management, and complications and other interventions; while outcomes on discharge were investigated using the modied Rankin scale (mRS). Descriptive statistics and multivariate logistic regression analyses were done. Root cause analysis was done to identify the causes of delay. patients. When considering the solutions to these challenges, the broader environment of practice ought to be taken into account.

the reasons for the delays in surgical intervention and poor outcomes in our patient population. By way of root cause analysis, the fractionation of identi ed problems into more manageable situations and scenarios allowed the recognition of speci c causes for such delays. Results:

Patient Demographics and Characteristics
During the speci ed time period, a total of 106 patients underwent surgical treatment of their ruptured aneurysm-majority of whom had favorable outcomes and cardiovascular comorbidities (Table 1). In terms of clinical grade upon presentation, the median scores were: Hunt and Hess grade 2, WFNS grade 2, and modi ed Fisher grade 3. Among all those who underwent surgical clipping, 75 (71%) patients had a good outcome (mRS 0-2) upon discharge. Fifteen of the surgically-managed patients died and overall mortality rate was 14.1%. The patients with unfavorable outcomes (mRS 3-6) had signi cantly higher median clinical grades on admission in terms of Hunt and Hess (p = 0.002) and WFNS grades (p < 0.001). Among their subgroup, majority had signi cantly higher proportion of preoperative rebleeding (16.13% versus 4%, p = 0.045) and intraoperative rupture (29% versus 2.67%, p < 0.001).

Time to Admission and Timing of Surgery
The median time interval from rst aneurysm rupture (day 0) to hospital admission were 5 days among those with favorable outcomes and 3 days among those with dependency or death ( Table 1). Regardless of outcome, a greater proportion of the patients underwent surgical clipping at a later time of more than 10 days from rst rupture, compared to those who underwent surgery within 3-10 days and those who had early clipping within 48 hours from aneurysm rupture. The median time interval from rst aneurysm rupture to surgery (range 1-302 days) were 14 days for those who had good outcomes and 16 days for those who had poor outcomes. In terms of timing of surgery, data were strati ed into early (0 to 2 days), intermediate (3)(4)(5)(6)(7)(8)(9)(10) days), or late (> 10 days) as per the landmark study by Kassell and colleagues. 33 Majority of the patients in this series underwent late surgery (67.0%) followed by intermediate time interval (29.2%); while only a minority (3.7%) had early intervention. In terms of both their time interval to admission and time interval to surgery, the differences between patients with good and poor outcomes were not signi cant.

Interventions and Complications
Those patients with unfavorable outcomes had a signi cantly higher proportion of having had the need for osmotic therapy which is the use of mannitol, and triple-H therapy which is the induction of hypertension, hemodilution, and hypervolemia. Only 11 (10.38%) of surgically-managed patients were admitted to the intensive care unit (ICU). There was no signi cant difference between patients with favorable versus unfavorable outcomes in terms of admission to the ICU, as well as various other medical and pharmacologic treatment. For those patients with unfavorable outcomes, there were signi cantly higher rates of vasospasm, increased intracranial pressure requiring CSF diversion or decompressive hemicraniectomy, healthcareassociated infections, and surgical or medical complications (Table 1).

Risk Factors for Dependency and Mortality
Simple logistic regression analysis of risk factors for unfavorable outcomes revealed the following covariates to be statistically signi cant: female sex (OR 2.79, 95% CI 1.14-6.86, p = 0.025), higher WFNS grades 3 (OR For the outcome of mortality (mRS 6) alone, exact logistic regression was done due to the relatively small proportion of patients who died. The following risk factors were found to be statistically signi cant ( When the timing of surgery was analyzed in terms of dependency (mRS ≥ 3) and mortality (mRS 6), the patients with poor outcomes did not differ signi cantly among the three treatment times (Table 3).
When controlling for patient demographics and other covariates, stepwise multivariate logistic regression analysis revealed three independent predictors for unfavorable outcomes (

Causes of Delays
Viewed through the lens of conceptual frameworks after root cause analysis of the thematic factors identi ed, we determined the causes can be classi ed in terms of being patient-related, provider-related, or healthsystems-related factors. Majority of patients who need neurosurgical intervention for this illness have geographic limitations and other barriers to surmount prior to their admission in a referral hospital with neurosurgical service. Financial barriers are present when costly diagnostics and other surgical implements require out-of-pocket expenses, or when catastrophic expenditures result in impoverishment. Misconceptions also abound from both patient and health-provider sides. These misconceptions may be sociocultural or nancial in context and are often about the risks and bene ts of the procedure, the care and referral pathways, or the risks and prognosis of the disease. Owing to the decentralized but disorganized nature of hospital referral systems in the country-between rural to urban / lower-level to higher-level centers-delays are magni ed by lack of hospital catchment arrangements (e.g. a neurosurgical center can refuse a transfer due to current high volume) and logistical concerns (e.g. lack of adequate ambulance transport to conduct the patient). On an institutional level, the hospital often has inconsistent availability of supplies (e.g. procurement of clips and other incidentals are delayed), and the ine ciency of operating room (OR) throughput is matched also by lack of related and needed infrastructure (e.g. bottleneck in OR queue of emergency procedures such as surgical clipping because of few surgical theatres or insu cient nursing workforce). On a health-system level, the lack of universal health insurance coverage for the whole population and the chronic underfunding of public hospitals with neurosurgical centers stand out as two important root causes of delays. All the identi ed causes are in turn in uenced by the broader environmental considerations of workforce, infrastructure and service delivery, and nancial and governance arrangements (Fig. 1).

Discussion:
The Quality Chasm Viewed Through the Lens of Neurosurgical Treatment of Aneurysmal Subarachnoid Hemorrhage In this series of socioeconomically disadvantaged patients who survived the initial rupture of SAH and were treated with upfront surgical clipping, we report the late timing of surgery experienced by majority of them, the factors predictive of their poor outcomes, and the causes of delay that are speci c in a low-resource setting. While many of these socioeconomic factors would be beyond the scope of usual clinical and neurosurgical practice, an awareness of the context allows the realization of the magnitude of systemic problems 10 that are more prevalent in LMICs. 11 From such knowledge, solutions could then be formulated. Although by no means exhaustive, we also proffer solutions broadly categorized into either micro-level interventions directed at clinical decision making or macro-level interventions directed at health systems. 11,12 The example of our series of patients and the institutional and country contexts will now serve as focal points for our discussion.

Deliberate Upfront Clipping in Low-Resource Settings
Since the publication of long-term outcomes from seminal trials, 13,14 patients with ruptured aneurysms ought to be offered the endovascular option for treatment. Emerging recommendations on standards of care, when situated in the context of nancial constraints at both the provider and receiver ends of neurosurgical care, pose a negative experience of contrasts. The trend in developed countries is a rise in the use of endovascular treatment option, or at the very least an involvement of a multidisciplinary team for reaching a consensus on the initial course of treatment. This is not the case however in LMICs, where the costs of coils and modern devices are prohibitive-and there are still few specialists who could offer this treatment option. 15 Providers of care regardless of specialty therefore commonly apprise the socioeconomically disadvantaged patients with the clipping option only. The socioeconomic context of our patient population does not constitute a justi cation to be accepted but rather becomes a re ection of the gap between current evidence and actual practice. Despite the realities, it is incumbent for decision makers, on behalf of the recipients of neurosurgical care whom they represent, to have the symmetry of information about the range of treatment options in order to provide the right informed consent. Expanding the workforce of endovascular neurosurgeons and neurointerventionalists-when coupled with systems-level public policy and use of competitive market forces 16 -also augurs well with increasing accessibility to this treatment option.

Late Timing of Surgical Clipping-A Symptom of Systemic Problems
Peculiar in our study setting would be the late time interval to surgery for the majority of the patients. Although the timing of surgery is not statistically signi cant-similar to what was found in studies with clipping as the exclusive intervention 17,18,19,20,21 -this remains a modi able factor in the institutional or regional catchment setup of our center, partly because the delays are avoidable for the most part.
Variations in surgical care occur on various levels; 8 the causes of which are mostly due to the locus of control of either the physician or the patient and relatives. Similar to the literature, several reasons for the delays were physician-related: physician diagnostic problems (37%) and delayed referral policy (23%) from the series of Kassell and colleagues, 22 sending patient home or wrong diagnosis for at least one day in the diagnosis of SAH (26%) from the series of Hernesniemi and colleagues. 23 Conversely, some of the delays in their series 22,23 were patient-related: unstable patient condition, failure of patients to recognize severity of illness, logistical reasons, and untimely decisions by patients and relatives. These reasons are also commonplace in our setting, but the added problem of scarce resources further magnify the causes of delay.
Other than the causes beyond the level of the individual patient, it becomes essential to identify which of those factors might be amenable to modi cation from the perspective of the provider. For example, we have taken measures to institute an open-line network communication channel 24 in order to improve inter-hospital coordination of transfers as well as pre-hospital management. Efforts have been made to address the timeliness of surgery by doing the surgical clipping beyond OR hours, even during the evenings, at which time the OR throughput is usually decreased. Often, when the workforce is adequate, it is the limitation of OR infrastructure that causes the bottleneck. Our data also illustrate the disparities from the expectations for timely and appropriate care of the patient with SAH in LMICs, 25 where reasonable timeframes are often unmet especially in our setting. Much work remains to be done in order to improve e ciency and expand the capacity of under-resourced neurosurgical centers in meeting the demands of the increasing surgical volume.
Aside from the fact that many patients from our center's catchment population live below the poverty threshold, the publicly-funded health system that is supposed to provide an enabling mechanism for them to receive proper neurosurgical care continues to be underfunded. 26 This twin problem at both receiving and providing ends of the spectrum remains the perennial limitation in publicly-funded neurosurgical centers such as our setting. Late timing of surgical treatment is also an outward manifestation of a structural problem-it represents the magnitude of the fragmentation of a publicly-funded healthcare system. The nancing of hospital-based inpatient services in the public setting are often limited by capitation-based reimbursements from the national social health insurance because the overall scarce resources of the national government are prioritized for population-based health programs. 27 From our example of neurosurgical service delivery, the undersupply and depletion of relevant medical supplies such as aneurysm clips, arterial line, bone ap implants, and other surgical implements highlight the ine ciencies inherent in the procurement processes and supply chain management of government-funded centers of any health system. 28,29,30 Improving systemslevel governance 31 and nancial arrangements 32 by streamlining how common neurosurgical supplies are purchased might improve access to care, and in turn health outcomes. 11 Given that hemorrhagic stroke, including SAH, is one of the leading causes of mortality in the country, 27 it becomes necessary to make essential neurosurgical care as one of the priorities for long-term capital expenditures and investments by the governing health sector. 11 In those patients waiting for their ruptured aneurysm to be clipped, the future risk of poor outcome needs to be understood with respect to both the impact of the cerebral insult they initially had during ictus and the arguably increased likelihood of them avoiding other complications like vasospasm, nosocomial infection, and many other time-dependent factors like rebleeding and even hospitalization costs-some of which may be curtailed or avoided had they undergone an earlier surgery. 33,34 While early aneurysm treatment remains to be recommended and is therefore preferred even in low-resource settings, this ideal still remains a daunting task to achieve.

Vasospasm and Intraoperative Rupture-Hardly Modi able Factors
Our data identify vasospasm as an independent predictor of mortality. Rosengart and colleagues 35 pointed out that while there are several studies that examined the factors associated with outcome prior to the era of widespread endovascular use, only 3 studies 36,37,38 have su cient study population to allow analysis of the effect of multiple independent factors. Our ndings are consistent with reports from those studies, that vasospasm is an enduring predictor of unfavorable outcomes. 35,37,38 Therefore, systems of care have to be responsive in diagnosing and treating this prophylactically and reversibly. This is not the case in our institution, where asymptomatic angiographic vasospasm may have been underdiagnosed and undertreated when compared to global state-of-the-art treatment, partly because of limitations of equipment and infrastructure, 3 and the lack of endovascular treatment options. 15 Notwithstanding the fact that neurointervention also grants the SAH patient with several options for the diagnosis and novel treatment of vasospasm-a risk factor predictive of mortality in this study-the increase in accessibility and availability of endovascular suites and care providers should ll a huge gap of the long unmet nationwide need for this niche of subspecialty care.
Our data also reveal intraoperative aneurysm rupture at the time of surgery as an independent predictor of mortality. Whereas other studies 39,40,41 only identify it as signi cant factor for poor outcome, Batjer and colleagues 41 went further to characterize it as largely dependent on speci c microsurgical techniques employed by the aneurysm surgeon. The rapidity of rupture control, which re ects the dexterity of the primary surgeon, may be similarly as important as intraoperative rupture. 39,42,43 At the time, the operating microscope for use during microneurosurgery was an old model (Fig. 2) that posed di culties in maneuverability for quick responses during rupture. Considering that technical pro ciency can be enhanced by judicious application of technology to operative imaging and visualization, equipment such as the operating microscope become invaluable tools that aid the neurosurgical procedures to become uneventful.
Regardless, our rates of unfavorable outcomes appear comparable to that of recent studies. 44,45 The rates of unfavorable outcomes were 43.3% and 22%, while mortality rates were 18.4% and 5.9% in studies by Roquer et al. 44 and Macdonald et al. 45 , respectively. Factors such as intraoperative rupture and vasospasm were similarly found to be independent predictors of poor outcomes in those studies, 44,45,46 despite the fact that their contemporary series of patients had been given the option of treatment by both open surgical and endovascular means. Our experience, however, remains to stand in stark contrast to that of ideal wellresourced settings. The care of the patient with SAH becomes especially di cult when standards of care cannot be offered-often due to inadequate infrastructure and even lack of supply of medicines. Suboptimal pharmacologic treatment often occurs when the supply of nimodipine runs out in the pharmacy, requiring outof-pocket expenditures from the family of the now unemployed breadwinner-patient with SAH. Other systemsrelated problems that are commonplace: a postop poor-grade patient who ideally requires ICU admission is instead cared for in the wards with a 1:20 nurse-to-patient ratio; or a neurologically deteriorating post-clipping patient who would have to be transported outside the hospital because the CT scanner has not been repaired for the longest time. These situations are testament to the fact that inadequate funding reduces the quality of care-and that addressing the problem of nite resources for neurosurgery in LMICs can have major effects on improving the outcomes of vulnerable patients. 11 The Way Forward Recognizing that social determinants of health 47 can affect the health of a population, a systems approach to the problem-solving of delays, can improve outcomes of interventions to a large extent. Advocacy work for public health and hospital improvement is also needed therefore, in order to lower the barriers to care can address health inequities, particularly in terms of public-private investments in infrastructure and equipment. 3,10,11 Preparing the needed workforce entails capacity-building and aligning of payment policies to serve the goal of quality and e ciency. All these must be entrenched in public policy and regulatory frameworks for sustainability and consistency, 27 because after all, improvements in care processes result in better quality.

Limitations of the Study
The study excluded patients with presumptive or con rmed diagnosis of aneurysmal SAH but who died before hospital admission or aneurysm treatment. In as far as assessing outcomes of the delayed treatment group, there is also a high likelihood of survival bias. Long-term outcomes after discharge as well as surrogate end points such as post-clipping durability of treatment were not assessed due to lack of postoperative angiography. Additionally, the correlations of our study variables may have been under-or overrepresented because of the retrospective study design. All these constitute a selection bias. Other confounding factors may also be present despite our exhaustive list of variables. For example, we did not have data for potentially clinically-signi cant variables like duration of brain retraction and temporary occlusion time, due to limitations of data recording. Other variables that have not been found to be statistically signi cant in this present study, and yet are ordinarily considered to be clinically signi cant in majority of cases of SAH, remain potential topics for future intensive investigation.
The system-level causes of delay identi ed in this study were not subjected to statistical analyses. Certainly they have an effect in one way or another, 11 but we did not demonstrate how these causes or even potential solutions may be strongly or weakly affective of outcomes. Studies that explore the link of nation-level indicators and indices to patient outcomes are recommended to ll this knowledge gap. We also hope that our experience of neurosurgical care in LMICs-particularly with regard to cerebrovascular treatment-can inform the process of guideline 3-6 formulation, in a manner similar to how the living guidelines for the management of traumatic brain injury take into account global experiences. 48 Further studies of high-quality evidence in LMIC settings are therefore needed.

Page 10/19
Conclusion: In terms of neurosurgical care for survivors of SAH in low-resource settings, the quality chasm-between the ideal and the reality-is indeed wide. From our experience of care for our series of socioeconomically disadvantaged patients, we provide the reasons for delays to seeking treatment and surgical clipping as they relate to vasospasm and intraoperative rupture-two independent predictors of both dependency and death in this series. Given that a favorable outcome is often precarious and not guaranteed for all patients suffering from both SAH and a low socioeconomic status, the way forward entails a focus on modi able causes that address both micro-level (provider-directed) and macro-level (health systems-directed) solutions.
Neurosurgeons who serve in under-resourced centers have the gargantuan task of systematically addressing the noted risk factors of unfavorable outcomes, while simultaneously needing to take health systems into account. The accessibility and availability of proper neurosurgical care in LMICs can be viewed as deeply disjointed from the global sense of how the quality of neurosurgical should be, thereby calling for much work that remains to be done in the task of bridging this chasm. Tables   Table 1. Univariate and logistic regression analysis of factors related to unfavorable outcomes in patients with SAH