Carpal Tunnel Syndrome as a Test Case for Value Assessment during the Pre-Surgical Period: The Impact of Structure and Processes of Care

Abstract

Background:

Few performance measures assess pre-surgical value (quality and utilization).

Objectives:

Using carpal tunnel syndrome (CTS) as a case study, 1) develop a model to evaluate pre-surgical quality and utilization and 2) identify opportunities for value improvement.

Research Design:

A retrospective cohort study utilizing Veterans Affairs (VA) national administrative data.

Subjects:

Patients who were evaluated in a VA primary care clinic on at least one occasion for CTS and received carpal tunnel release (CTR) over a 7-year period.

Measures:

We modeled facility-level performance on two outcomes: surgical delay (marker of quality) and number of pre-surgical encounters (utilization) for CTS, and examined association between patient, facility, and care process variables and performance.

Results:

Among 41,912 Veterans undergoing CTR at 127 VA medical centers, the median facility-level predicted probability of surgical delay was 48%, with 16 (13%) facilities having significantly less delay than the median and 13 (10%) facilities having greater delay. The median facility-level predicted number of pre-surgical encounters was 8.8 visits, with 22 (17%) facilities having significantly fewer encounters and 22 (17%) facilities having more. Care processes had a stronger association with both outcomes than structural variables included in the models. Processes associated with the greatest deviations in predicted delay and utilization included receipt of repeat electrodiagnostic testing, use of 2 or more nonoperative treatments, and community referral outside of VA.

Conclusions:

Using CTS as a test case, this study demonstrates the potential to assess pre-surgical value and identify modifiable care processes associated with pre-surgical delay and utilization performance.

Introduction

Current healthcare reforms seek to improve quality and care coordination by “bundling” payments for post-surgical care over an extended time period. However, little attention has been paid to measuring value in pre-surgical care, which would include one or more measures of quality, resource utilization, or both. During this pre-surgical period, patients often have multiple touches with the healthcare system, which is an important opportunity to identify ways to improve care coordination, quality, and timely access.

Treatment for carpal tunnel syndrome (CTS) is an ideal context for developing measurement models of care quality between primary care and specialty providers. CTS is common, with more than 500,000 carpal tunnel release (CTR) operations performed in the U.S. annually.¹ While patients with new and mild symptoms can often be treated conservatively, delays in definitive treatment among patients with persistent or severe symptoms can lead to worse outcomes and prolonged disability,^2–6 reflecting the importance of timely, efficient, and appropriate pre-surgical care. Thus, CTS is an ideal condition to develop a model to measure pre-surgical quality and utilization. Such a model might guide efforts to identify systems-level strategies to improve care coordination, access, and patient outcomes more broadly. Although surgical outcomes rightly deserve our attention, little is known about aspects that influence quality and utilization during pre-surgical care.

Therefore, we sought to develop an episode-based approach to evaluate a measure of pre-surgical care value (quality and utilization) relevant to the health system and patient perspective, using patients with CTS as a case study. We further sought to assess facility variation in the pre-surgical performance metrics and the impact of modifiable structure and processes of care on pre-surgical quality and utilization, to potentially identify future targets for quality and value improvement. The Veterans Affairs (VA) healthcare system, the largest integrated system in the United States, provides an excellent opportunity to assess such measures, given its ability to follow patients nationally and longitudinally.

Methods

Data Collection and Cohort Inclusion

The study cohort included Veterans who were evaluated in a VA primary care clinic on at least one occasion for CTS and ultimately underwent their first CTR surgery between January 1, 2010 and December 31, 2016.

The primary data source included domains available from the VA Corporate Data Warehouse (CDW). The study cohort included Veterans receiving a unilateral or bilateral CTR operation (identified by CPT codes 29848, 64721) in an outpatient setting. To avoid confounding due to concurrent hand conditions or complicated CTS, patients were excluded if they received additional major procedures at the same time as the CTR operation or if they received CTR on the same or contralateral extremity in the 10 years prior. In addition, Veterans designated as having an emergency American Society of Anesthesiologists (ASA) status were excluded. Finally, as the primary research aim was to measure pre-surgical care, patients without receipt of at least 1 primary care visit within the 2 years prior to the index CTR surgery were excluded from the cohort to ensure an established relationship with a VA primary care clinician.

Pre-Surgical Episodes of Care

We created initial and recurrent pre-surgical episodes of care (Appendix A) for cohort patients using a model based on the National Quality Forum patient-centered episodes of care theoretical construct⁷ coupled with Donabedian’s structure-process-outcomes framework.⁸ All CTS-related services that patients received starting with the initial encounter for CTS-related symptoms up to the first CTR surgery were abstracted from the CDW Outpatient and Fee-Basis (Community Care) Domains, including use of CTS-related diagnostic tests, non-operative treatments, specialty care visits, and referral for community (see Appendix B for diagnosis and procedure codes). The inclusion period was chosen to end in 2016 given concern about reliably capturing community care use within the pre-surgical episodes after 2016 due to transition of VA purchased care (community care) information to a new data system. Care that patients received within 12 months of their earliest presentation was defined as the initial pre-surgical episode. Patients who continued to seek medical care for CTS-related symptoms after the initial pre-surgical episode were defined as having persistent CTS symptoms. The first visit after the initial 12-month treatment period was defined as the start of a recurrent pre-surgical episode. The duration of the recurrent episode ended at the time of surgery given that all patients in the study cohort received surgical release.

Pre-Surgical Performance Measures

CTS-related care between the initial presentation and surgery was examined to create two performance measures that would impact pre-surgical value (quality and utilization). Surgical delay was chosen as a measure of pre-surgical quality and access. The number of pre-surgical encounters for CTS-related care was chosen as a measure of pre-surgical utilization.

Surgical Delay (Quality)

There is strong evidence that surgery has greater long-term benefit (at 6- and 12-months follow-up) and improved outcomes compared to nonoperative therapies for CTS treatment.^9–15 For these reasons, surgical delay was conservatively defined when the duration of the recurrent episode was greater than 6 months (Appendix A). This conservative definition of delay also gives providers the benefit of the doubt if patients initially had a mild presentation during the initial episode, and factors in the potential for preference-related delays in initiating care for ongoing symptoms.

Pre-Surgical Encounters (Utilization)

The total number of encounters were calculated as a measure of utilization during the entire pre-surgical period. Relevant encounters included clinic, therapy, and diagnostic testing visits for CTS or related symptoms, including the first CTS-related visit until surgery. Encounters were categorized according to the clinician specialty, including primary care and specialty clinics (occupational therapy, physical medicine and rehabilitation, pain, neurology, and surgery). Diagnostic testing was categorized by type including: X-ray, ultrasound, CT, MRI, and electrodiagnostic studies (EDS) (Appendix B).

Independent Variables

Patient-Level Variables

Patient age, sex, race, and VA priority group comprised the patient-level demographic characteristics recorded. VA priority groups represent 8 categories of service-related disability and economic need. Factors related to patient health status were also recorded, including number of Charlson Comorbidity Index comorbidities (Deyo adaption)^16,17, diabetes, renal failure, and ASA status at the time of CTR. High-risk medications that could impact delay or utilization were recorded, including receipt of anti-coagulants, anti-platelet medications, and oral steroids.

Facility-Level Variables

Patients were assigned two facility-level identifiers based on their specific site of primary care and specific site of CTR surgery. CTR facility assignment for patients receiving surgery in the community was assigned based on the facility assuming payment responsibility. The selection of structural variables was informed by the Institute of Medicine’s Model of Access to Personal Health Care Services¹⁸ which outlined the number, type, concentration, location, and organizational configuration of healthcare providers as potential structural barriers to care. Consequently, we constructed several structural variables that could be derived from VA and community care administrative data to approximate access to CTR, including specialist/surgeon workforce availability (CTR facility surgical specialist full-time equivalent, annual volume of CTR surgeries relative to facility patient volume, proximity of specialty providers to patients and primary care clinics, and primary care facility type (e.g., Medical center versus primary care community-based clinics).

Process Variables

We constructed process variables based on treatment factors evaluated in our preliminary studies^19–21 or clinical experience that could potentially impact delay and utilization, including the use and timing of imaging and EDS (pre-surgeon referral, post-referral, both pre- and post-referral, and no testing), use of inappropriate imaging (CT, MRI, or ultrasound), use of and total number of non-operative therapies (including splints, occupational therapy, steroid injection, and oral steroids for CTS) (see Appendix B for list of CPT/HCPCS codes), and referral to nonsurgical specialists (pain, rheumatology, neurology, and physical medicine and rehabilitation). Use of oral steroids were considered CTS treatment if the prescription was filled 0–7 days after a CTS-related encounter, and excluded patients taking oral steroids greater than 90 days for any reason and patients with diagnoses indicative of chronic disease in which steroids may be prescribed (see Appendix B).

Statistical Modelling

We performed regression models using a fully-Bayesian framework²² to handle the relative complexity of the models and provide flexibility in the estimation tasks. Surgical delay (recurrent episode >6 months duration) was modeled through logistic regression. The number of discrete encounters for CTS-related pre-surgical care was modelled with a negative-binomial distribution. Weakly Informative priors were used in all models as the data set was quite large, thus the contribution to the posterior by the prior distributions was imperceptible. Empty (intercept-only) three-level hierarchical models for surgical delay and pre-surgical utilization were first created to estimate reliability-adjusted facility-level performance, with the patient’s specific PCP location nested within the patient’s CTR surgery facility. Analyses were conducted at the heath system (PCP site nested within CTR site) given the challenge of assigning responsibility of care processes to a single provider or specialty. Separate CTR facility-level performance ranks were created based on the probability of surgical delay and the number of pre-surgical encounters predicted from the respective models. Finally, models evaluating the relationship of delay and utilization outcomes to structure and process variables were computed incorporating the patient, structure, and process independent variables described above. The width of all credible intervals was set to 95%, chosen for convention. Clinical significance was defined as non-overlap of 95% credible intervals (from Bayesian analysis) when assessing differences between two or more estimates. Data analyses were performed using the statistical programming language R (v3.6.0) and the brms package (v2.12.0).

Results

Patient-Level Surgical Delay and Utilization Outcomes

The final study cohort included 41,912 patients who underwent their first CTR surgery during the study period (Figure 1). The cohort was predominantly Caucasian males with a mean age of 57 years (SD:12.6) (Appendix C). Surgical delay occurred in 47% (n=19,862) of the cohort, with a median duration of recurrent episodes of 16 months (IQR:5–51). The cohort had a median of number of 7 pre-surgical visits (IQR:5–11). The median time between the index (or initial) CTS-related encounter and CTR was 2.4 years (Interquartile range [IQR]:0.7–6.5), over which the number of pre-surgical visits was assessed.

Figure 1.

Study cohort inclusion and exclusion flow diagram.

Facility-Level Surgical Delay and Utilization Outcomes

Among 127 VA medical centers performing CTR, the median facility-level rate of surgical delay was 48%, with a wide range of facility delay rates ranging from 0 to 84%. There was a median of 7 pre-surgical encounters among each facility-level median utilization, with facility-level medians ranging from 1 to 14 pre-surgical visits. Figure 2 depicts the reliability-adjusted estimates of surgical delay and utilization for each medical center performing CTR. Overall, the median facility-level predicted probability of surgical-delay was 48% (IQR:44–53%), with 16 (13%) facilities performing significantly better than the IQR (less delay), and 13 (10%) medical centers performing significantly worse (greater delay). The median facility-level predicted number of pre-surgical encounters was 8.8 (IQR:8.1–10.0), with 22 (17%) facilities performing significantly better than the IQR (fewer visits), and 22 (17%) facilities performing significantly worse (greater visits).

Figure 2a and 2b.

Facility-level estimates (n=127) of probability of surgical delay and utilization with 95% credible interval for each medical center. Shaded regions represent estimates falling between the 25th and 75th percentile of facility-level probability of delay (0.44–0.53) and facility-level pre-surgical visits (8.1–10.0), respectively. Dashed line represents 50th percentile (median) of facility-level estimates (48% of patients experiencing delay and 8.8 visits).

Multivariable Models of Surgical Delay and Utilization

Descriptive statistics of the care process and structural variables are outlined in Table 1. Overall, care processes had a stronger association to delay and utilization outcomes than the structural variables included in the two models (Tables 2 and 3). Care processes that resulted in the greatest deviation in predicted delay relative to the population mean of 47% included receipt of repeat diagnostic testing (both before and after surgeon consultation), use of 2 or more nonoperative treatments, and community referral outside of VA (Table 2). These same care processes along with having an encounter with a nonsurgical specialist similarly resulted in the greatest deviations in predicted pre-surgical utilization relative to each reference comparison group (Table 3).

Table 1.

Structure and Process Variables

Patient-Level Structure Variables	N	%
Patient primary care clinic facility type
VA medical center	29,811	71.1
Health care center	1,384	3.3
Primary care community-based outpatient clinic	4,347	10.4
Multi-specialty community-based outpatient clinic	4,698	11.2
Other	273	0.7
Unclassified	1,399	3.3
Patient access to multiple providers in same building
Primary care and surgical specialist	29,666	70.8
Primary care and nonsurgical specialist	32,623	77.8
Primary care and occupational or physical therapy	34,029	81.2
Surgery facility complexity
High	34,139	81.5
Moderate	5,300	12.6
Low	2,435	5.8
Unspecified	37	0.1
Facility-Level Structure Variables	N	Mean (SD)
Facility-level surgical specialist clinical full-time equivalents	130	3.3 (2.6)
Facility-level CTR annual volume per 10K core unique patients	127	62 (46)
Patient-Level Process Variables	N	%
Encounter with nonsurgical specialist	24,531	58.5
Encounter for electrodiagnostic study	38,417	91.7
Timing of electrodiagnostic study
Before surgical consult	26,226	62.6
After surgical consult	5,423	12.9
Both before and after surgical consult (repeat testing)	3,985	9.5
No study performed	3,495	8.3
Indeterminant	2,783	6.6
Encounter for imaging	13,842	33.0
Use of inappropriate imaging	1,362	3.2
Timing of first imaging study
Before surgical consult	5,941	14.2
After surgical consult	5,990	14.3
Both before and after surgical consult (repeat testing)	896	2.1
No study performed	28,070	67.0
Indeterminant	1,015	2.4
Use of steroid injection	5,683	13.6
Use of splints	24,074	57.4
Use of specific therapeutic modalities	10,601	25.3
Use of oral steroids for carpal tunnel	939	2.2
Total number of categories of nonoperative treatments
0	15,015	35.8
1	14,712	35.1
2	10,059	24.0
3	2,037	4.9
4	89	0.2
Community care referral for carpal tunnel-related care	5,702	13.60

Table 2.

Multivariable Regression of Surgical Delay Outcome ^*

	OR	95% CI	Delay Predicted Probability	95% CI
Structural Components
Primary care clinic facility type
VA medical center	reference group		0.48	0.47 – 0.50
Health care center	1.03	0.87 – 1.22	0.49	0.45 – 0.54
Primary care community-based outpatient clinic	1.14	1.04 – 1.24	0.51	0.49 – 0.54
Multi-specialty community-based outpatient clinic	1.09	1.00 – 1.18	0.50	0.48 – 0.53
Other	1.38	1.01 – 1.84	0.56	0.49 – 0.63
Unclassified	0.94	0.82 – 1.07	0.47	0.43 – 0.51
Access to PCP and surgical specialist in same building
Yes	1.09	1.00 – 1.20	0.49	0.48 – 0.51
No	reference group		0.47	0.45 – 0.50
Access to PCP and non-surgical specialist in same building
Yes	1.07	0.98 – 1.18	0.49	0.47 – 0.51
No	reference group		0.47	0.45 – 0.50
Access to PCP and physical therapist in same building
Yes	1.12	1.02 – 1.24	0.49	0.48 – 0.51
No	reference group		0.47	0.44 – 0.49
Surgery facility complexity
High	reference group		0.49	0.47 – 0.51
Medium	1.00	0.84 – 1.19	0.48	0.43 – 0.53
Low	0.98	0.78 – 1.21	0.49	0.45 – 0.53
Facility-level surgical specialist clinical full-time equivalents **	0.91	0.83 – 1.01	0.48	0.45 – 0.50
Facility-level CTR annual volume per 10,000 core unique patients **	0.98	0.83 – 1.16	0.50	0.46 – 0.54
Process Variables
Encounter with nonsurgical specialist
Yes	1.72	1.62 – 1.82	0.54	0.53 – 0.56
No	reference group		0.41	0.39 – 0.43
Timing of electrodiagnostic study
Before surgical consult	reference group		0.45	0.43 – 0.47
After surgical consult	1.54	1.44 – 1.65	0.56	0.56 – 0.58
Both before and after surgical consult (repeat testing)	4.90	4.46 – 5.41	0.80	0.78 – 0.82
No study performed	0.69	0.63 – 0.76	0.36	0.34 – 0.39
Indeterminant	0.68	0.60 – 0.76	0.36	0.33 – 0.39
Timing of first imaging study
Before surgical consult	1.41	1.32 – 1.50	0.54	0.52 – 0.57
After surgical consult	1.30	1.21 – 1.39	0.53	0.50 – 0.55
Both before and after surgical consult (repeat testing)	2.66	2.21 – 3.15	0.69	0.65 – 0.73
No study performed	reference group		0.46	0.44 – 0.49
Indeterminant	1.36	1.13 – 1.61	0.53	0.49 – 0.58
Use of Inappropriate Imaging
Yes	1.28	1.11 – 1.48	0.55	0.51 – 0.59
No	reference group		0.49	0.47 – 0.50
Total number of categories of nonoperative treatments
0	reference group		0.36	0.34 – 0.38
1	1.79	1.70 – 1.89	0.50	0.48 – 0.52
2	2.94	2.77 – 3.13	0.62	0.60 – 0.64
3	4.57	4.01 – 5.16	0.72	0.69 – 0.74
4	4.13	2.11 – 8.87	0.69	0.53 – 0.82
Community care referral for carpal tunnel-related care
Yes	1.79	1.65 – 1.95	0.61	0.59 – 0.64
No	reference group		0.47	0.45 – 0.49

^*Model also controlled for the following patient demographic characteristics and risk factors: age, sex, race, VA priority group, ASA status, diabetes, renal failure, use of anticoagulation or antiplatelet medications at time of surgery, use of oral steroids between initial carpal tunnel encounter and surgery

^**Post-estimation predicted probabilities for continuous independent variables were centered and scaled around population mean and standard deviation for each variable

Abbreviations: ASA, American Society of Anesthesiologists; CI, credible interval; CTR, carpal tunnel release; OR, odds ratio; PCP, Primary care provider

Table 3.

Multivariable Regression of Pre-Surgical Utilization Outcome

	Number of Pre-Surgical Encounters
	IRR	95% CI
Structural Components
Primary care clinic facility type
VA medical center	reference
Health care center	1.06	1.02 – 1.11
Primary care community-based outpatient clinic	1.11	1.08 – 1.13
Multi-specialty community-based outpatient clinic	1.13	1.05 – 1.21
Other	1.10	1.08 – 1.12
Unclassified	0.95	0.92 – 0.98
Access to PCP and surgical specialist in same building
Yes	1.03	1.01 – 1.06
No	reference
Access to PCP and non-surgical specialist in same building
Yes	1.05	1.02 – 1.08
No	reference
Access to PCP and physical therapist in same building
Yes	1.03	1.00 – 1.06
No	reference
Surgery facility complexity
High	reference
Medium	0.95	0.88 – 1.02
Low	0.96	0.90 – 1.02
Facility-level surgical specialist clinical full-time equivalents	0.94	0.92 – 0.97
Facility-level CTR annual volume per 10K core unique patients	0.91	0.85 – 0.96
Process Variables
Encounter with nonsurgical specialist
Yes	1.32	1.31 – 1.34
No	reference
Timing of electrodiagnostic study
Before surgical consult	reference
After surgical consult	1.16	1.15 – 1.18
Both before and after surgical consult (repeat testing)	1.50	1.47 – 1.52
No study performed	0.83	0.81 – 0.84
Indeterminant	0.77	0.75 – 0.79
Timing of first imaging study
Before surgical consult	1.13	1.11 – 1.15
After surgical consult	1.13	1.12 – 1.15
Both before and after surgical consult (repeat testing)	1.31	1.27 – 1.35
No study performed	reference
Indeterminant	1.14	1.10 – 1.19
Use of Inappropriate Imaging
Yes	1.00	0.97 – 1.02
No	reference
Total number of categories of nonoperative treatments
0	reference
1	1.27	1.26 – 1.29
2	1.59	1.56 – 1.61
3	1.93	1.88 – 1.97
4	2.32	2.12 – 2.53
Community care referral for carpal tunnel-related care
Yes	1.31	1.28 – 1.33
No	reference

^*Model also controlled for the following patient demographic characteristics and risk factors: age, sex, race, VA priority group, ASA status, diabetes, renal failure, use of anticoagulation or antiplatelet medications at time of surgery, use of oral steroids between initial carpal tunnel encounter and surgery

Abbreviations: ASA, American Society of Anesthesiologists; CI, credible interval; CTR, carpal tunnel release; IRR, incidence rate ratio; PCP, Primary care provider

Timing of Electrodiagnostic Services and Imaging

91% (n=26,226) of patients received EDS in the pre-surgical period, with most (63%, n=26,226) receiving testing before the surgical consultation, and 9.5% (n=3,985) receiving EDS both before and after surgery consultation (Table 1). The predicted probability of surgical delay increased substantially with successive degree of EDS use, with predicted delay of 36% for those not receiving EDS at all (lower than the 47% population average), 45% predicted delay for those having EDS prior to surgery consultation, 56% delay probability for those receiving testing after surgical consultation, and 80% predicted delay probability if EDS was performed before and after surgery consultation (Table 2). Similar patterns were seen in the relationship between the number of pre-surgical encounters and EDS use (Table 3).

Imaging during the pre-surgical period was performed in 33% (n=13,842) of the cohort (Table 1). Inappropriate imaging (CT, MRI, or US for CTS diagnosis) occurred in 3% of patients (n=1,362). Receipt of any imaging test was associated with greater surgical delay probability and increased utilization compare to patients not receiving imaging (Tables 2 and 3).

Non-Operative Treatments

Splints were the most common type of nonoperative treatment used (57%; n=24,074), followed by use of specific therapeutic modalities (25%; n=10,601), steroid injection (14%; n=5,683), and oral steroids (2.2%; n=939). Among these types of nonoperative treatments, 29% of patients (n=12,185) used two or more (Table 1). Patients experienced an increase in predicted probability of surgical delay relative to the population average as the number of nonsurgical treatments increased, with 62% predicted delay probability for patients receiving 2 types of nonoperative treatment, and 72% delay probability for patients using three types (Table 2). A similar relationship was seen between an increasing number of nonoperative treatments and number of pre-surgical encounters (Table 3).

Community Care Use

Of the 41,912 patients sampled, 14% were referred from the VA to a community clinician for CTS-related care. Community care use was associated with a 14% increase in the predicted surgical delay probability relative to patients who received all care in the VA (61% and 47% predicted delay probability, respectively) (Table 2). Pre-surgical utilization demonstrated a similar trend, where predicted pre-surgical utilization was 31% higher among patients using community care compared to patients receiving all CTS-related care in the VA (Table 3).

Discussion

This study demonstrates the potential to use an episode-based approach to assess pre-surgical quality and utilization, and to identify care processes to target for future quality improvement, using patients receiving surgery for CTS as a test case. This evaluation demonstrated quality problems (surgical delay associated with low value diagnostic testing) and excess resource use (multiple pre-surgical visits) that occur during pre-surgical CTS care, which may not be appreciated with existing quality and utilization evaluation models. There was considerable variation in facility performance in predicted surgical delay and pre-surgical utilization, with 10% of VA medical centers performing significantly worse than the median VA facility, and 17% of VA medical centers averaged significantly more pre-surgical visits than the median facility. Furthermore, we identified modifiable processes associated with greater delay in CTS and excess utilization, including use of multiple nonsurgical treatments, use of repeat testing before and after surgeon referral, and community referral outside of the VA. In contrast, facility structural attributes were not as strongly associated with delays or excess utilization.

Inclusion of pre-surgical care within the surgical episode could help to identify systems and processes in need of quality improvement.²³ In widely-used care episode models, such as the Centers for Medicare and Medicaid Bundled Payment for Care Improvement Advanced (BPCI Advanced), episodes begin at the time of surgery or admission and includes services up to 90 days after discharge.^24,25 These programs ignore the quality, utilization, and efficiency of care that occur between an initial encounter for a surgical problem and receipt of surgery. In the current test case, the pre-surgical episode measures of quality and utilization not only demonstrated systems-level variability in pre-surgical quality and utilization, but the models were also used to identify modifiable care processes that could be targeted to improve pre-surgical quality and efficiency. Reducing surgical delay and improving efficiency of pre-surgical care is important for patients with CTS given previous studies that demonstrated worse outcomes after CTR with prolonged use of nonoperative treatments and longer lasting symptoms compared to patients having earlier surgery.^2–6

Through the findings of this study, health systems could reduce surgical delay and improve efficiency by reducing repeat testing and avoiding use of more than one type of nonoperative treatment when an initial treatment fails. The 2017 Royal College of Surgeons’ Commissioning Guide for Treatment of Carpal Tunnel Syndrome suggests that failure of one conservative treatment is a predictor that others will also fail.²⁶ Recognizing the likelihood of treatment failure with subsequent nonoperative treatments along with the associated delay and higher utilization may help to improve patient outcomes with more timely surgical care.

The routine use of electrodiagnostic studies (EDS) is another debated care process for CTS patients that deserves greater attention. Studies are consistent in finding that EDS is not a reference standard to diagnose CTS and that physical exam alone is sufficient to diagnose CTS in patients with classic symptoms.^27–31 In patients with a high probability of CTS based on history and physical exam criteria alone, EDS does little to change the probability of diagnosis.^30,32 EDS has also been previously shown in the private sector to prolong time to surgery, which is consistent with our study’s findings.¹⁹ More than 90% of patients received EDS prior to CTR in our cohort of VA patients, whereas 58% of patients undergoing CTR in the private sector received EDS as assessed in a previous study of national administrative data.¹⁹ The reasons for this higher rate of EDS use in the VA is unclear. However, patients who avoided EDS in this study had the lowest predicted probability of surgical delay (36%), and patients who received testing both before and after surgeon consultation (repeat testing) had the greatest probability of delay (80%). Given that it is a common practice for many surgeons to request EDS prior to an initial evaluation for CTS,²¹ surgical delay and utilization may be reduced by identifying patients who would not benefit from testing and by avoiding repeat testing when possible.

This study also demonstrates the need for greater attention to the timeliness of surgical care for patients referred outside of the VA to community providers. As a result of the 2014 Veterans Choice Act and 2018 Maintaining Internal Systems and Strengthening Integrated Outside Networks (MISSION) Act, a greater number of Veteran appointments are scheduled in the community. As a result of the MISSION Act, roughly 640,000 additional Veteran patients are estimated to be referred to the community each year at an average cost of $8,600 per patient.³³ Much of the emphasis on expanding access to community providers has been focused on reducing wait times for an initial appointment.³⁴ Given the increased delay and utilization associated with community care in this study, as well as prolonged time between PCP referral and surgery reported in our prior work³⁵ among patients receiving a combination of VA and community care, greater attention should be made in identifying and rectifying sources of delay among patients referred to the community.

This study has several limitations inherent to the use of administrative data. We were unable to determine the impact of clinical severity and patient preferences on time to surgery and utilization. Thus, at an individual level, reasons for surgical delay and intensity of utilization were unknown. For this reason, we sought to compare differences at the health system level to understand high level differences across VA facilities and the association of patient, clinical, and systems factors with the quality and utilization outcomes. The VA integrated health system lends itself to measure longitudinal episodes of care in various healthcare settings and at a national level, which is more difficult to accomplish reliably among patients in the private sector. Thus, the degree that these findings are generalizable to the private sector are unknown. Lastly, the study cohort only included patients with CTS who received surgery given prior work that demonstrated greater facility-level variation in nonoperative service use in patients receiving surgery (compared to patients receiving only nonoperative treatment)²⁰ and to reduce clinical severity heterogeneity as much as feasible with administrative data. We hypothesize that processes that improve efficiency of care among surgical patients will also improve quality and efficiency of care for patients receiving only nonoperative treatment. However, care trajectories of patients treated only nonoperatively relative to surgical patients is outside the scope of this study and deserves future attention.

Despite these limitations, our study highlights potential benefits of evaluating pre-surgical care in the potential to improve the value of surgical care more broadly and to identify targets for future quality improvement efforts. A mixed-method approach will be necessary, as an extension of this study, to understand the nature of modifiable processes in need of improvement before attributing performance to a single provider or specialty. The evaluation of quality and efficiency of pre-surgical care is critical in situations in which outcomes are dependent on care timeliness and in clinical scenarios that would benefit from reduction of low-value pre-surgical tests and treatments. Expansion of episode-based measurements of quality and utilization to include pre-surgical care has the potential to improve outcomes and care efficiency for patients in this CTS test case scenario. A similar approach could be tailored to other conditions in which timeliness of care has the potential to impact postoperative outcomes and recovery.