Shunt infusion studies: impact on patient outcome, including health economics

We identified and collected the results of all shunt infusion tests performed between January 2013 and December 2015. These included a total of 280 computerised CSF infusion tests which were performed in 210 shunted hydrocephalus patients of all ages and aetiologies. As part of our routine clinical pathway, all shunted individuals presenting with sudden onset, clear symptoms of raised ICP and an unequivocal CT head would have had an urgent revision and therefore do not require a shunt infusion test.

Besides the hydrocephalus shunt infusions, we included 85 tests on PTCS patients (including idiopathic intracranial hypertension (IIH)) and analysed them separately because of differences and more complexities compared to hydrocephalus patients. However, the shunts implanted in those patients do not differ significantly in properties and the same principles of shunt testing in vivo apply to them as well, as infusion objectively tests the hydrodynamic properties of the implanted shunt.

Infusion test results and conclusion on the shunt’s function had been reported independently by a clinical physicist at the time of request and the conclusion statement (blocked, under-/overdraining and properly functioning) was used to match against clinical outcome.

Objective testing of implanted shunts through infusion studies is a well-refined method based on both in vitro and in vivo knowledge of all marketed shunts’ hydrodynamics, primarily the shunt critical pressure and its resistance [6, 10, 13]. Validation of the methodology in relationship to clinical and intraoperative findings of shunt revision has recently been performed in a large paediatric population from two European centres [17].

Analytically, testing for underdrainage relies on the calculation of mainly the critical shunt pressure, with the baseline pressure very occasionally being elevated in relationship to the shunt operating pressure. The critical shunt pressure is calculated based on the shunt resistance, infusion rate, operating pressure and average normal abdominal pressure of 5 mmHg as critical ICP (in mmHg) ≤ resistance of shunt × infusion rate + shunt operating pressure + 5 (abdominal pressure). An illustrative example of underdrainage is shown in Fig. 1a. In relationship to overdrainage, this can easily be assessed by testing the siphoning effect of the shunt when the patient is sitting upright, as well as assessing the difference between baseline ICP and ICP after performing the tilt test (upright position for at least 10 min until stabilisation of pressure, then flat position again). A pictorial depiction of this is shown in Fig. 1b. When the plateau pressure during infusion of liquid and the shunt resistance are significantly higher than the manufactured properties, combined with a higher than the manufactured resistance and excluding increased intra-abdominal pressure, a distal obstruction (valve and/or distal catheter) can be suspected. The baseline ICP is usually normal, especially in cases of chronic hydrocephalus. Occasionally, and depending on the aetiology and acuity of hydrocephalus, the baseline ICP can be significantly raised to levels too high for a functioning shunt, and a distal obstruction can be confirmed right away. Representative examples of such distal obstructions are shown in Fig. 2a, b. A ventricular catheter obstruction is accordingly observed in the absence of a detectable ICP pulse waveform when accessing the ventricular space through insertion of needles to the reservoir or prechamber. A classic example of proximal obstruction is shown in Fig. 2c. Analysis can be highly diagnostic and may also suggest partial obstruction, slit ventricles, etc. [6‐9, 11, 15, 23, 26].

We carefully inspected all follow-up documentation from the patients’ records through our local electronic medical records (e-MR) and Epic softwares up to 6 months as well as up to 12 months from the time of the test. Searching included all follow-up clinics, possible readmissions, revision surgeries, inpatient notes if admitted and all reasons for the above. Based on the documented clinical progress and consultant impression, and mainly on the basis of relief of symptoms with good reported quality of life at follow-up, patients were marked as improving or non-improving.

Patients with various forms of hydrocephalus are discussed separately from patients with PTCS (including IIH) in the outcomes analysis to reflect differences in their clinical course and management.

13/155 tests were linked to further care needs postinfusion: 4 patients required revision before a 6-month follow-up period. Three out of these 4 were most probably independent of the time and the results of the study. They required revisions because of an accident that exposed their shunt tubing (one case) or because of distal catheter migration in the other two patients a few months after the test. Three patients were lost in follow-up and three more died of causes that could not be associated with intracranial hypertension or shunt-related complications; these were for each of the three: hip fracture, recurrence of aneurysmal bleeding and a complex epilepsy syndrome. Lastly, three patients with a most likely patent shunt had ongoing symptoms and were referred for discussion in the Multidisciplinary Team meeting.

142/155 tests resulted in cases that were either discharged from neurosurgical care with instructions or referred for headache management. In some occasions, they had their shunt setting adjusted and did not require any further neurosurgical intervention. Five of these patients needed a longer period of supervision and management, mostly trying to adjust their shunts to a setting that relieved their symptoms but were managed successfully without surgery. Two patients remained unwell; one had never been well before or after shunting for NPH and had already previously suffered from severe overdrainage, requiring evacuation of bilateral subdural haematomas. The other patient appeared to suffer from refractory headaches despite attempts of adjusting their shunt setting; however, they were still managed in an outpatient basis.

Overall, 140 of the tests revealed no evident shunt malfunction on 109 unique hydrocephalic patients (27 individuals had 2 and 2 patients had 3 tests, each negative) who improved with conservative management at 6 months of follow-up.

Similarly, 136 cases out of the 140 that had done well at 6 months remained well after for at least 12 months, with the exception of 3 patients who required revision within 12–14 months after their initial infusion test, and one who required a revision after approximately 7 months (the single patient with refractory headaches that could not improve at 6 months by setting manipulation). Three more patients were referred for a new infusion test that showed a problem with the shunt that required changing of the setting and no revision, their follow-up records showing improvement.

The reasons for revision in all these patients are illustrated in Table 1.

Overall, 43 revision operations were performed in patients with evidence of obstruction on infusion studies; 2 revisions as well as 2 infusions were performed in 2 patients who required revision and their new shunts blocked shortly after surgery. For 33 of them, the infusion study had indicated blockage (proximal, distal or both catheters), 8 underdrainage and 2 overdrainage.

Thirty-two improved sustainably at 6 months and even after 1 year, 5 by changing their shunt setting in the meantime. For some of the rest, it could be possible that their setting was changed but it was not noted in our records and some of them were discharged or referred to neurology.

The remaining 11 patients had no change in their symptoms, with persisting headaches dominating in all of them. Six had to be discussed in our MDT meeting for possible styloidectomy, stenting, etc. Unfortunately, 5 showed no improvement even after their shunt setting was adjusted postsurgically (for a long time of follow-up, 6–12 months, even after neurology referral).

The reasons for revision in this group of patients are illustrated in Table 1.

Of the other 82/125, 1 was lost in follow-up. Four patients were classified as non-improving. Two presented with new episodes of seizures and had to be managed with antiepileptics and the other 2 were discussed in a MDT meeting ([16, 24]).

Seventy-seven non-revised patients with conservative management (setting manipulation or discharge/neurology referral) had improvement or acceptable control of their symptoms and did not require additional care for at least 1 year after their initial encounter for infusion studies. Two of these cases had evidence of proximal shunt obstruction but clinical indications of improvement soon after the infusion study, sustained for at least a year, indicating resolution of the obstruction, most likely from flushing the proximal catheter during infusion. One also had radiological evidence of significantly smaller ventricles.

From the 81/82 with non-obstructive shunt malfunction and available follow-up and the 149/155 with normal shunt function available for follow-up, a total of 142 + 77 = 219/230 clinically showed that they did not require revision and were symptom-free. This translates to a 95.22% negative predictive value (NPV) of shunt testing in vivo for excluding shunt obstruction.

Four hydrocephalus patients were admitted electively after the infusion test for overnight ICP monitoring. Two of the monitoring sessions confirmed overdrainage (there was suspicion but not strong evidence of overdrainage in the infusion test) and two required assessment of ICP and it dynamics for clinical and safety reasons, because a proximal obstruction did not allow reading and recording of the ICP during the infusion test.

Five PTCS patients were referred for ICP monitoring due to unresolving symptoms suspicious of active disease.

Our infection rate both for acute meningitis and chronic, subacute infections was 0%. This was assessed through patients’ medical records and revision requirements. All infections and CSF samples related to shunts are referred to Cambridge regardless of local area, from all of our catchment area on East of England.

The cost of day case admission for patients with possible shunt malfunction were calculated by the Trust’s Finance Department and included:

The total cost of each reservoir infusion study to the Trust was 844 GBP that matched the exact income received.

Each shunt revision is billed to reflect

The total cost of a shunt revision procedure ranged from 9437 to 12,436 GBP (average of 10,937 GBP). The wide range in costing is due to different comorbidities that affected the cost of the surgery, as well as perioperative complications, care needs and length of stay.

From our outcome analysis data, it can be extrapolated that

In respect of what might have happened in the absence of a CSF infusion test service, two scenarios have been explored: first, that all patients would have had a shunt revision in the absence of a CSF infusion study, OR second, that a proportion of patients admitted for observation would have had ICP monitoring, 70% of whom would have gone on to shunt revision (decision trees—Fig. 5a, b).

Scenario 1

Scenario 2

Therefore, from all of the above, the total saving to the Trust by providing a CSF infusion service = ~ £2,766,081 over 3 years = ~ £922,027 per annum.

When comparing infusion studies to a standard protocol, using only ICP monitoring and exploratory surgery, the overall financial benefit could be approximately £442,710 per 100 patients admitted with possible shunt malfunction. An analytical decision tree showing a comparative cost analysis between using infusion studies versus no infusion studies is presented in Fig. 5.

Infusion tests appeared quite accurate and efficient in directing away from shunt malfunction in hydrocephalus, as demonstrated by the 212 cases that were managed conservatively without further hospitalisations. Shunt testing in vivo appears to be most valuable to our selected cases that did not have a radiological signs substantiative of shunt malfunction and were not acutely unwell. This can often be the case with NPH patients, as well as younger patients with chronic or neglected hydrocephalus [2, 6, 8, 22, 23, 25, 43].

It has been highlighted in multiple reports of paediatric hydrocephalus patients that an increased number of revision operations is a negative predictive factor for cognitive outcome and overall disability and quality of life for paediatric patients [3‐6, 9, 27, 28]. For adults, this increased number also is a predictor of further shunt revision surgery requirements in the future and therefore constant care needs, many hospitalisations and reduced quality of life [3, 6, 17, 29]. Importantly, over 85% of our patients lead a symptom-free and neurosurgically uneventful life for at least 1 year of follow-up. Furthermore, patients were spared the consequences of longer hospital stay and the additional complications of unnecessary shunt revision surgery, whose rates and sequelae have been reported extensively in the literature [18, 19, 27].

The pathophysiology of PTCS is complex and often involves concomitant elevation of the venous sinus pressures and CSF pressures that appeared to be coupled [12, 33, 34, 40]. However, a shunt infusion could be useful in pointing towards or away from a shunt problem, facilitating patient flow and referral for stenting or other management. There were 15 patients with proven working shunts on their CSF infusion studies whose symptoms could not be relieved with conservative management. They were referred back to the PTCS multidisciplinary team for further investigations and consideration of alternative treatments without being subjected to a revision that would most likely not have benefitted them and would have delayed their referral to a more appropriate service.

Based on the data from symptom relief and follow-up alone, the NPV of shunt infusion tests was 95%. This refers to the accuracy with which a shunt infusion test can exclude a shunt obstruction. In order to accurately reflect the positive predictive value as well as the NPV of the test, the results from the intraoperative shunt flow testing should be compared with the infusion test result. As our patients with a negative (exclusive of obstruction) shunt infusion test would not undergo a revision in their majority, it would not be possible to calculate this in our current dataset. Furthermore, as from our 51 revised patients, not all the intraoperative notes clearly indicated the site of obstruction; the low number of patients currently tested would not be able to yield a reliable positive predictive value. Finally, we only calculated the NPV derived from hydrocephalus patients, whose clinical course as mentioned is more straightforward, whereas PTCS patients present with multidisciplinary challenges that confound their assessment based on shunt patency alone. A new design, aiming to test this diagnostic accuracy with a high number of revised patients, should be performed to address this.

Overall, the surgical management of CSF disorders is very cost-effective in comparison with the rest of medicine in general. The average cost per QALY was only £215 in 1990 in comparison, for example, to £750 for a hip replacement or £700 for a pacemaker. Length of stay has reduced considerably since 1990 so that the cost per QALY will have reduced further in relative terms. Antibiotic-impregnated shunt catheters have been shown to be cost-effective with cost savings of $42,125 and $230,390 per 100 de novo shunts placed in adult and paediatric patients respectively [35].

As expected, avoiding surgical revisions in shunted patients seems to be of considerable financial benefit to the NHS. However, as stated in the methods, it is not possible to know exactly how many of these patients would have been selected for shunt revision, if infusion studies were not available. There are other methods that are used worldwide in order to diagnose shunt malfunction. Overnight ICP monitoring remains the gold standard, but it is also quite invasive, involves cranial surgery and has its own cost implications. Other methods, less invasive, involve radiation (such as radio-contrast shuntograms, radioactive flow studies) or no radiation (MRI/high resolution MRI, optic sheath diameter measurements, etc.). However, they all implicate the cost of a radiologist or other highly specialised clinical staff, expose the patient to radiation and even endanger the functionality and/or patency of the shunt [2, 4, 6, 10, 23]. In addition, measurements of steady-state ICP cannot exclude shunt malfunction. No flow through a shunt does not automatically mean obstruction, but many other things, including inadequate pressure to open the valve or even collapsed ventricles (slit ventricles) around the proximal catheter, among others [4, 6, 10, 23, 28, 29, 32, 43]. Last but not least, our current protocol for performing shunt infusion studies that includes strict aseptic technique and proper cleaning and disinfection appears to be 100% effective in avoiding additional costs related to infections from the procedure.

Shunt revision is a predictive factor for multiple future revisions; therefore, averting one redundant revision translates to averting multiple revisions in the long-term [3, 4, 7‐9, 39]. A recent report from the UK shunt registry has indicated the revision rates in all UK centres, where our centre holds a lower revision rate compared to the average, especially after first implantation [41]. Another aspect considered in the revision costs, is that the comprehensive nature of infusion studies allows the clinical team to be aware, and not blinded to, the location of the shunt issue. A proximal catheter obstruction can easily be differentiated from a valve or distal catheter problem [6, 7, 10, 11, 23, 26, 28, 32, 38]. This most probably decreases the cost of the revision surgery, since parts of the shunt can be left intact, decreasing surgical time, use of equipment, complications and postoperative hospital stay. However, investigating this was not within the scope of our current paper and could be the subject of a different study.