Comparing Microvascular Outcomes at a Large Integrated Health Maintenance Organization with Flagship Centers in the United States

Home , Sinai (Noguchi)

ORIGINAL ARTICLE Comparing Microvascular Outcomes at a Large Integrated Health Maintenance Organization With Flagship Centers in the United States

Edgar A. Lueg, MD, FRCSC

Objective: To determine if patients undergoing micro- Results: All 5 flagship centers are major academic health vascular reconstructive head and neck surgery (MRHNS) centers ranked in the top 18 “best head and neck hospi- at a large integrated health maintenance organization can tals” in the United States. Flap loss (1.7% vs 4.4% for flag- expect outcomes similar to some of the best or flagship ship centers; range, 0.9%-8.8%) and mortality (2.6% vs 2.8% centers in the United States. for flagship centers; range, 0.5%-6.3%) rates were not significantly different. Although lengths of stay in flagship cen- Design: Outcomes (flap loss, mortality, length of stay), ters were similar to each other and the literature (mean, 21.4 days; range, 20.1-22.5 days), our length of stay was eligibility (recent consecutive US center experience), high- Ͻ experience (100 cases), high-volume (26 cases per year), significantly shorter (8.8 days, P .001). and flagship criteria were prospectively defined. A sys- Conclusion: For high-experience and high-volume cen- tematic MEDLINE search identified 17 eligible reports. ters, patients undergoing MRHNS at a large integrated Independent, blinded medical reviewers identified 5 cen- health maintenance organization can expect morbidity ters (29%) as flagship centers. and mortality outcomes similar to flagship centers in the United States, with shorter hospitalizations. Patients: The first 116 consecutive patients (average, 39 cases per year) who underwent MRHNS on this service. Arch Otolaryngol Head Neck Surg. 2004;130:779-785

HE QUALITY OF THE TER- METHODS tiary surgery provided by major academic health cen- INTEGRATED HMO SERIES ters (MAHCs)1 in the United States is recog- Roughly 1 in every 4 insured Southern Cali- nizedT worldwide. On the other hand, Kai- fornians are members of Kaiser-Permanente, ser-Permanente, the nation’s largest inte- a large, nonprofit, group-model (patients grated (physicians and hospitals share choose their physicians from our group), in- financial risk2) health maintenance orga- tegrated HMO, which provides comprehen- nization (HMO), now provides most ter- sive health care.2 The Southern California tiary surgery for its own members. In 1998, Permanente Medical Group is composed of a dedicated microvascular reconstructive more than 4000 multispecialty (partner and af- filiated) physicians who are exclusively con- head and neck surgery (MRHNS) service tracted to provide their health care at roughly was established to provide this for our 200 outpatient facilities and 12 major medi- members living in the Los Angeles (Calif) cal centers across Southern California. I per- referral basin (south of Fresno and north formed all 117 free flaps (Table 2), most on of San Diego) who underwent oncologic our dedicated Flap Tuesday, with the other resection at our Regional Head, Neck, and head and neck surgical oncologist (HNSO) (Mi- Skullbase Surgical Oncology Center. This chael P. McNicoll, MD) assisting. All patients From the Microvascular service was the first high-volume (see flag- were evaluated prospectively at our multidis- Reconstructive Head and Neck ship criteria) MRHNS service to be estab- ciplinary (HNSOs and radiation oncologists) Surgery Service, Regional Head, lished outside an MAHC in the United tumor board and received extensive preopera- Neck, and Skullbase Surgical tive medical and psychological preparation. As Table 1 3-7 Oncology Center, Southern States ( ). The objective of this team leader, the patient’s admitting HNSO man- California Permanente Medical study was to determine if patients under- aged the surgical issues. Medical issues were Group, Los Angeles. The author going MRHNS at a large integrated HMO managed by an intensivist in the dedicated in- has no relevant financial can expect outcomes similar to some of the tensive care unit and a hospitalist on the dedi- interest in this article. best or flagship centers in the United States. cated ward. Discharge planners coordinated

(REPRINTED) ARCH OTOLARYNGOL HEAD NECK SURG/ VOL 130, JUNE 2004 WWW.ARCHOTO.COM 779

Variable Description Integrated HMO Flagship Mean Flagship 1 Flagship 2 Flagship 3 Flagship 4 Flagship 5 Series median patient (year)* 1999 1992 1991 1993 1990 1997 1989 Series duration, y 3.0 6.7 3.8 13.0 5.0 2.8 9.0 Total patients, No. 116 314 288 698 184 115 286 Total flaps, No. 117 332 308 728 200 119 305 Center volume, patients per year, No. 39 48 76 54 37 41 32 Surgeon volume, No. of MRHNS procedures performed 39 25 19 18 37 41 11 Early (first 100) flaps, % 85† 43† 32 14 50 84 33

Abbreviations: HMO, health maintenance organization; MRHNS, microvascular reconstructive head and neck surgery. *Calendar year during which half of the series’ patients had already undergone surgery. †Significant by Newcombe-Wilson hybrid score 95% confidence interval of difference.

Table 2. Surgery Comparison

Integrated HMO, Flagship Mean Significant Flagship 1, Flagship 2, Flagship 3, Flagship 4, Flagship 5, Description No. (%) (95% CI)* Difference No. (%) No. (%) No. (%) No. (%) No. (%) Sample size (flaps) 117 (100) 332 NA 308 (100) 728 (100) 200 (100) 119 (100) 305 (100) Basic techniques Secondary cases 4 (3) 11 (−0.16 to 0.00) No 43 (14) 14 (2) 29 (14) 5 (4) 68 (22) Nonworkhorse donor sites 4 (3) 11 (−0.16 to 0.00) No 67 (22) 56 (8) 18 (9) 6 (5) 39 (13) Vein grafted cases 0 (0) 4 (−0.10 to 0.00) No 17 (6) 7 (1) 16 (8) 0 20 (7) Simultaneous donor sites 1 (1) 4 (−0.09 to 0.02) No NR 7 (1) 11 (6) 1 (1) 18 (6) Composite (bone) cases 30 (26) 33 (−0.19 to 0.06) No 91 (30) 221 (30) 120 (60) 29 (24) 70 (23) Donor site selection Radial forearm 64 (55) 31 (0.10 to 0.36) Yes 79 (26) 218 (30) 47 (24) 69 (58) 57 (19) Fibular 27 (23) 15 (−0.03 to 0.19) No 53 (17) 200 (27) 7 (4) 27 (23) 15 (5) Anterolateral thigh 22 (19) 0 (0.12 to 0.28) Yes 00000 Extremity total† 113 (97) 48 (0.38 to 0.59) Yes 138 (45) 427 (59) 57 (28) 96 (81) 80 (26) Rectus abdominis 1 (1) 16 (−0.23 to −0.08) Yes 44 (14) 173 (24) 20 (10) 17 (14) 48 (16) Latissimus dorsi 0 (0) 5 (−0.07 to −0.02) Yes 23 (7) 19 (3) 2 (1) 3 (3) 40 (13) Scapular/parascapular 3 (3) 6 (−0.10 to 0.03) No 19 (6) 21 (3) 19 (10) 1 (1) 32 (10) Jejunum 0 (0) 13 (−0.21 to −0.07) Yes 65 (21) 81 (11) 5 (2) 0 89 (29) Iliac crest 0 (0) 15 (−0.23 to −0.08) Yes 19 (6) NR 96 (48) 2 (2) 11 (4) Torso total† 4 (3) 52 (−0.59 to −0.38) Yes 170 (55) 301 (41) 143 (72) 23 (19) 225 (74)

Abbreviations: CI, confidence interval; HMO, health maintenance organization; NA, not applicable; NR, not reported. *Difference between integrated HMO and flagship mean calculated by Newcombe-Wilson hybrid method. †Totals include miscellaneous sites not listed separately in this table.

outpatient resources, including routine daily home nursing vis- FLAGSHIP CENTER SELECTION its. After the first 57 patients, all the following patients were placed on a target 6-day inpatient clinical care pathway (CCP) Similar (single US center, complete, consecutive, MRHNS expe- (Figure 1).8 The study period began with the first service case rience reports of cases not restricted to a particular defect or do- on October 27, 1998, and was closed on the anniversary after nor site) and recent (published within 10 years of the close of this having met the anticipated high-experience criteria. There- study) reports met the eligibility criteria. Flagship centers were fore, the first 116 consecutive patients (average, 39 cases per defined as those centers that have published an eligible report that year) who underwent MRHNS on this service between Octo- also satisfies both the high-experience (at least 100 consecutive ber 27, 1998, and October 27, 2001 were the subjects of this cases) and high-volume (averages at least 1 case every few weeks comparison (Table 3). or 26 cases per year) criteria. High-experience centers were selected, since most microvascular surgeons believe that greater ex- MAIN OUTCOME SELECTION perience is associated with improved outcomes.3-7,9,11-15 High- volume centers were selected, since a “broad body of evidence Flap loss (patients with complete flap loss) was selected because from large, population-based studies” has consistently demon- it is the most commonly used specific measure of reconstructive strated better outcomes for both major oncologic resections and morbidity following MRHNS3-7 and most microvascular sur- high-risk vascular procedures.10 Once the outcomes, eligibility, geons believe it is an indicator (albeit relatively crude) of qual- high-experience, high-volume, and flagship criteria were specifi- ity.9 Perioperative mortality (any cause within 30 days) was se- cally defined, a systematic MEDLINE search was undertaken. A lected because it is the most commonly used measure of mortality review of the 722 citations containing the words free flap following MRHNS3-7 and most health care experts believe it is an and head within their text identified 17 eligible reports. indicator of quality.10 Length of stay (LOS) (complete contigu- Two independent (no interest in the findings of this study) and ous days) was selected because it is the most commonly used mea- blinded (no knowledge of the study’s objective) medical review- sure of hospitalization following MRHNS5,11-15 and most health ers evaluated the abstracts (text only with center identifiers re- care experts believe it is an indicator of resource cost.16 moved) for the flagship criteria. Both independently excluded

(REPRINTED) ARCH OTOLARYNGOL HEAD NECK SURG/ VOL 130, JUNE 2004 WWW.ARCHOTO.COM 780

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 the same 12 reports. Therefore, 5 centers (29%) were defined A flagship (Table 1). (51-Day Stay Truncated at UCL) 24 UCL = 23.2 d 21 PROCESS CONTROL CHART Zone A 18 We used a process control chart (PCC)8 to graph our LOS to ana- 15 Zone B CCP Patients 12 lyze hospitalization (Figure 1). A PCC is simply a standardized Zone C LOS, d statistical graph of a key quality characteristic (LOS in our case) 9 Mean = 8.8 d of a given process over time. They have been used by the manu- 6 Zone C 6 d 3 facturing sector for a long time to help decide whether a pro- Zone B 0 LCL = 0 cess is stable (only common, random, or normal variation is pres- 0 57 113 ent) or unstable (a special cause of variation is also present). This Consecutive Discharged Patients, No. decision is important because common-cause variation pro- duces a normal (standard bell-shaped) distribution and is there- B fore predictable (within statistical limits). On the other hand, the 18 decision that a process is unstable often leads to an investiga- 15 CCP Patients 12 tion into the source of the special-cause variation. More re- 9 UCL = 9.5 d cently, the usefulness of PCCs in helping to achieve continuous 6 8 d Variation, 3 Mean = 2.6 d quality improvement in health care has been recognized. Up- 0 LCL = 0 per and lower confidence limits are the absolute boundaries out- 0 57 113 side which a particular occurrence will be arbitrarily consid- Consecutive Discharged Patients, No. ered indicative of special-cause variation. These boundaries are conventionally set at 3 SDs from the mean to minimize the com- Figure 1. Decreasing length and variation in the hospitalization process.8 bined risk of mistakenly considering a truly common-cause oc- Process control chart (see the “Methods” section) of 113 consecutive discharged patients’ length of stay (LOS) (A) and its variation (B).8 Dashed currence to be indicative of special-cause variation (type I er- lines represent upper confidence limits (UCLs) and lower confidence limits ror, which may lead to tampering with a stable process) and visa (LCLs), the boundaries of normal variation. Dotted lines separate probability versa (type II error, which may lead to undercontrolling an un- zones (full standard deviations) labeled C, B, and A, moving away from the stable process). It is extremely unlikely that a common-cause oc- mean. CCP indicates clinical care pathway. currence will lie outside these boundaries (type I error, 0.27%). The areas between these boundaries are also conventionally di- portional differences (all except mean age) for this series and vided into 3 equal probability zones (labeled C, B, and A, mov- the flagship mean were calculated by the Newcombe-Wilson ing away from the mean), which coincide with whole standard hybrid score method.20 Differences whose 95% CI did not en- deviations. These zones are useful in identifying special-cause compass a zero difference were considered statistically signifi- variation even if the occurrences lie within the upper and lower cant. Significance for the difference in mean age was tested us- confidence limits. For example, 15 consecutive occurrences within ing a random-effects, 1-way analysis of variance test using 1 SD of the mean (zone C) is also extremely unlikely to occur if standard deviations conservatively estimated from the sample the process is stable. ranges.21 STATISTICAL ANALYSIS RESULTS The 95% confidence intervals (CIs) for flap loss and mortality were determined based on experience that they more closely MAIN OUTCOMES approximate a binomial distribution.17 Significance was tested 1 with a Poisson heterogeneity test (a goodness-of-fit ␹2 test based All 5 flagship centers are MAHCs. In 2003, US News & on expected values from a weighted average Poisson model).18 World Report placed all these flagship centers among The 95% CIs for LOS were determined based on experience that America’s 18 “best hospitals for head and neck disor- it more closely approximates a normal distribution.11 In addi- ders.”22 Four are also ranked among the 12 “best cancer tion to our overall LOS, the CIs for our sickest (American So- hospitals,” including the 2 highest ranked National Can- Ͼ ciety of Anesthesiologists [ASA] class 2) and irradiated sub- cer Institute (NCI)–designated comprehensive cancer cen- groups were also determined to address concerns that these ters in the nation.22,23 Flap loss (1.7% vs a flagship aver- potential confounders might explain the significant differ- age of 4.4%; range, 0.9%-8.8%; P = .24) and perioperative ences in LOS. Since standard deviations for the flagship samples were not reported, we estimated them by multiplying the stan- mortality (2.6% vs a flagship of average 2.8%; range, 0.5%- dard deviation of the largest (145 discharges) cohort from a flag- 6.3%; P=.43) rates were not significantly different. These ship center reported19 by the ratio of the flagship’s average over findings were confirmed by the overlap between the 95% this cohort’s average. Based on past experience with other LOS CIs for our flap loss and mortality with the flagship cen- distributions, this scaling method usually provides a reason- ters (Figure 2). Although flagship LOSs were similar able, yet conservative, estimate of standard deviation (Raoul to each other and the literature (average, 21.4 days; range, Burchette, MS, written communication, October 31, 2002). Sig- 20.1-22.5 days),11 our overall (8.8 days, PϽ.001), sick- nificance was tested with an independent-groups, unequal vari- est (9.2 days, PϽ.001), irradiated (9.6 days, PϽ.001), and ance t test. Outcomes were extracted from the defining flag- Ͻ 3-7 pathway (7.1 days, P .001) LOSs were all significantly ship report. In several cases, LOS was not reported. In these shorter. These findings were confirmed by the wide sepa- cases, it was extracted from the largest flagship cohort reported.12-14 Since these samples were smaller, their CIs were ration between their and the flagship centers’ 95% CIs wider. All surgery and patient variables that were reliably ex- (Figure 3). Our PCC (Figure 1) demonstrates that our tractable (either specifically stated or unambiguously deriv- hospitalization process is unstable (special-cause varia- able) from at least 2 defining flagship reports were compared tion is present). All 56 consecutive LOSs after imple- directly with this series (Tables 2 and 3). The 95% CIs for pro- mentation of our CCP (57th discharge) is within 1 SD

(REPRINTED) ARCH OTOLARYNGOL HEAD NECK SURG/ VOL 130, JUNE 2004 WWW.ARCHOTO.COM 781

Integrated HMO, Flagship Mean Significant Flagship 1, Flagship 2, Flagship 3, Flagship 4, Flagship 5, Variable No. (%) (95% CI)* Difference No. (%) No. (%) No. (%) No. (%) No. (%) Sample size (patients) 116 (100) 314 NA 288 (100) 698 (100) 184 (100) 115 (100) 286 (100) Demographics Age, y Mean 60 (NA) 55 (NA) Yes† 55 (NA) 54 (NA) 58 (NA) 54‡ (NA) 55 (NA) Youngest 10 (NA) 11 (NA) NA 14 (NA) 3 (NA) 18 (NA) 19 (NA) 1 (NA) Oldest 95 (NA) 87 (NA) NA 89 (NA) 89 (NA) 80 (NA) 88 (NA) 87 (NA) Male sex 71 (61) 63 (−0.15 to 0.11) No 178 (62) 412 (59) 120 (65) 82 (71) 169 (59) Health status Sickest (ASA class Ͼ2) 61 (53) 53 (−0.14 to 0.14) No 212 (74) NR NR 37 (32) NR Tumor status Tumor Malignant 111 (96) 92 (−0.03 to 0.11) No 288 (100) NR 147 (80) 111 (97) 254 (89) SCC (% of malignant) 105 (95) 81 (0.05 to 0.23) Yes 223 (77) NR 129 (88) NR 196 (77) Site§ Aerodigestive (mucosal) 97 (83) 79 (−0.07 to 0.15) No 251 (81) 491 (67) 178 (89) 108 (91) 197 (65) Maxilla/orbit/skull base 13 (11) 18 (−0.17 to 0.03) No 34 (11) 226 (31) 15 (8) 11 (9) 94 (31) Cutaneous 7 (6) 4 (−0.05 to 0.09) No 23 (7) 11 (2) 7 (3) 0 14 (5) Large size (AJCC T-stage 3/4)࿣ 100 (97) 79 (0.09 to 0.27) Yes 94 (72) NR NR 85 (74) 158 (90) Prior radiation 14 (12) 36 (−0.35 to −0.12) Yes 87 (30) 239 (34) 79 (43) 39 (34) 113 (40)

Abbreviations: AJCC, American Joint Committee on Cancer; ASA, American Society of Anesthesiologists; CI, confidence interval; HMO, health maintenance organization; NA, not applicable; NR, not reported; SCC, squamous cell carcinoma. *Difference between integrated HMO and flagship mean calculated by Newcombe-Wilson hybrid method. †PϽ.001 by random effects 1-way analysis of variance calculation. ‡Estimate (midrange value) for random effects 1-way analysis of variance calculation. §Denominator for sites is flaps. ࿣Denominator for size is AJCC classifiable primary tumors.

POTENTIAL CENTER CONFOUNDERS A 13.2 Better surgical techniques generally improve outcomes over time. By the very nature of this study, our out- 8.8 comes are more recent (Table 1). To minimize this in- herent limitation, only recent (published within 10 years) Flagship reports were eligible. Although a 5-year cutoff would have

Flap Loss, % 4.4 Average diminished this effect even further, only 2 flagship centers (40%) would have remained. The M. D. Anderson 0 Cancer Center, the top-ranked NCI-designated compre- Integrated Flagship Flagship Flagship Flagship Flagship HMO 1 2 3 4 5 hensive cancer center in the nation, would have been 1 of those 3 centers excluded. However, the major im- B provements (microscope, instruments, flaps, training) in 8.4 MRHNS techniques were already available before most (more than 90%) flagship patients underwent their surgery.3-7 Furthermore, greater center and surgeon expe- 5.6 rience also improves MRHNS outcomes over time.3-7 Cer- tainly, our own outcomes, as well as the flagship centers’, Flagship

Mortality, % 2.8 Average are better today. However, most of the flagship series include a significantly smaller percentage of early (first 100) Ͻ 0 flaps (85% vs 43% for the flagship centers, P .05), which Integrated Flagship Flagship Flagship Flagship Flagship might have skewed our outcomes adversely.3-7 HMO 1 2 3 4 5 Still, flagship patients underwent surgery on aver- Figure 2. Morbidity and mortality outcomes. Comparison of 95% confidence age 6.3 years earlier, and there has been a recent trend intervals (error bars) for flap loss (A) and mortality (B) for this series and the nationally (especially in California) to decrease LOS.16 flagship centers. HMO indicates health maintenance organization. One flagship center has already reported a significant re- duction for a later cohort.13,19 It is very likely that the av- of the mean (zone C). Remember, even 15 consecutive erage LOS at all the flagship centers is significantly shorter occurrences within 1 SD are extremely unlikely to oc- today. However, our own LOS since implementation of cur if the process is stable (see the “Methods” section).8 a target 6-day inpatient CCP is now less than 1 week with The variation graph confirms the significant variation concurrent improvements in a wide array of safety (mor- dampening that occurred after implementation of the CCP bidity and mortality) outcomes (Figure 3).24 A system- (the main source of the special-cause variation). atic re-review of the original 722 MEDLINE citations con-

(REPRINTED) ARCH OTOLARYNGOL HEAD NECK SURG/ VOL 130, JUNE 2004 WWW.ARCHOTO.COM 782

Flagship 20 Average

10 Length of Stay, d Length of Stay, 5

0 Overall Sickest Irradiated Pathway Flagship Flagship Flagship Flagship 1 2 3 5

Figure 3. Hospitalization outcome. Comparison of 95% confidence intervals (error bars) for length of stay for this series’ cohorts and the flagship centers.

taining the text words free flap and head found the shortest that the flagship patients, although chronologically average LOS for a large (at least 100 cases) cohort of younger, were still sicker. MRHNS patients to be 11.2 days.11 Our CCP patients’ (now Of the disease (morbidity) or tumor status vari- more than 150 patients) LOS is significantly shorter than ables extractable from at least 2 defining flagship re- this (6.3 vs 11.2 days, PϽ.001).24 ports, malignant tumors (percentage of squamous cell car- cinomas [SCCs]), size (percentage with American Joint POTENTIAL PATIENT CONFOUNDERS Committee on Cancer [AJCC] primary tumor stage 3 or 4), and irradiated were significantly different (Table 3). Of the demographic variables extractable from at least 2 Unfortunately, the upper aerodigestive tract sites were defining flagship reports (Table 3),3-7 mean age was sig- partitioned inconsistently (ie, oropharynx was grouped nificantly different. Indicators of socioeconomic status with the oral cavity in some flagship reports and phar- (income, education, employment, neighborhood, in- ynx in others), limiting reliable extraction of compa- sured, minorities), which are often adversely associated rable mucosal subsite data. The malignant tumors in our with health status (comorbidities) and hence postop- patients were significantly more likely to be SCCs (95% erative morbidity and mortality, were not reported. A vs 81% for the flagship centers, PϽ.05). The difference single flagship center reported its percentage of non- seems to reflect a higher-than-expected contribution of whites, which was significantly higher for this series unusual tumor findings (sarcomas, basal cell carcino- (62% vs 18% for the flagship center, PϽ.05).4 Our pa- mas, non-SCCs) at 2 flagship centers (23% and 23%). The tients were also significantly older (5 years) on average AJCC classifiable SCC primary tumors in our patients were (60.2 vs 55.2 years for the flagship centers, PϽ. 001). also significantly more likely to be large (97% vs 79% in This difference reflects our conviction that one’s the flagship centers, PϽ.05). This difference reflects the chronologic status (age) on its own is not an indepen- fact that we are the designated regional tertiary-referral dent (directly) associated risk factor. This is under- surgical oncology center with a dedicated MRHNS sur- scored by the fact that we report the first MRHNS case geon. One flagship report3 addressed pathology and site in a nonagenarian patient (vigorous 95-year-old dis- and found no association with flap loss. To my knowl- charged on day 6).3-7 However, advanced chronologic edge, no report has addressed size (or stage), since most status is consistently associated with poorer health and MRHNS patients have large (advanced) primary tu- hence has been dependently (indirectly) associated mors. However, I know of no microvascular surgeon who with greater postoperative morbidity,3,5,6,7,15 mortal- believes that patients with large SCC tumors might have ity,3,5,6 and LOS15,24 in MRHNS patients. more favorable outcomes compared with patients with Most experienced microvascular surgeons would smaller non-SCC tumors. agree that a patient’s health status is the most powerful Significantly fewer of our patients underwent prior independent predictor of morbidity, mortality, and LOS irradiation (12% vs 36% for the flagship centers, PϽ.05). following MRHNS.3-7,11,15,25 Unfortunately, only 2 flag- This reflects the fact that our prospective multidisci- ship centers reported any direct measure of it, the rela- plinary tumor board generally recommends primary on- tively crude and nonspecific ASA class (Table 3). Our per- cologic resection followed by adjuvant radiation for ad- centage of sicker patients (ASA class Ͼ2) fell between vanced (AJCC clinical stage III or IV) resectable tumors. the 2 reported (53% vs 32% and 74% for the flagship cen- Therefore, we perform MRHNS on relatively few patients ters, PϾ.05).4,6 However, even if we completely exclude in whom irradiation fails, since in most cases aggressive our healthier patients, the 95% CI for the LOS of our sick- oncologic resection has also failed. However, the 95% CI est (ASA class Ͼ2) patients, all with severe or life- for the LOS of our irradiated patients remains widely sepa- threatening systemic disease, would still remain widely rated from the flagship centers’ CIs (Figure 3). Further- separated from the flagship CIs (Figure 3). Yet this does more, there are now at least 9 large, peer-reviewed not entirely eliminate the possibility that a more sensi- MEDLINE reports, 7 from flagship centers, involving al- tive and specific measure of health status for MRHNS pa- most 5000 free flaps that have specifically addressed the tients, such as the Washington University Head and Neck prior irradiation issue, and none have found that it was a Comorbidity Index (our mean was 0.54),26 might reveal risk factor for any of our outcomes.3-5,7,11,27-30

(REPRINTED) ARCH OTOLARYNGOL HEAD NECK SURG/ VOL 130, JUNE 2004 WWW.ARCHOTO.COM 783

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 POTENTIAL SURGERY CONFOUNDERS HMO and MAHC patient populations are different.1,2 Some of these differences may be potential confounders that This series reflects our basic techniques (Table 2). Like skewed our outcomes favorably or adversely (see the “Re- most of the flagship centers we avoid (if possible) sec- sults” section). Yet our outcomes were only selected in- ondary (delayed) reconstructions, nonworkhorse (infre- dicators of quality and cost.8-11,16 For example, although quently selected) donor sites, use of vein grafts, and mortality may be similar for 2 groups of patients, most sur- planned simultaneous (multiple) donor site reconstruc- geons would agree that quality was better if one group was tions. Excluding one statistical outlier (Ͼ3 SDs from the sicker before surgery.8,10,18 This study does not permit fair mean), our reported composite (bone flap) reconstruc- conclusions (and none are made) regarding quality or cost, tion rate was similar to the other flagship centers (26% since several likely confounders could not be adequately vs 27% for the flagship centers; range, 23%-30%; PϾ.05). controlled (statistically adjusted)18 with the flagship in- Our workhorse armamentarium (frequently [at least 10%] formation available. selected donor sites) includes the radial forearm, fibu- On the other hand, the size (243% longer), strength lar, and anterolateral thigh. All 3 have long, large vascu- (widely separated CIs), and stability (separation re- lar pedicles that have helped us avoid the need for vein mains even for sickest and irradiated strata) of the dif- grafts entirely. Use of the infrequently selected but mul- ference in LOS all suggest that it is robust. However, analy- tipaddle subscapular composite torso donor site3 has also sis of the flagship LOS based on extrapolation of our helped us entirely avoid the need for planned multido- subgroup variation for 3 important risk factors (poorer nor site reconstructions and the significant additional health status, large tumor status, and prior radiation sta- functional and aesthetic morbidity they inflict. tus) suggest that measured differences in case mix do not Where we do differ significantly with the flagship appear to adequately explain the large observed differ- centers is our avoidance of torso donor sites (3% vs 52% ences in LOS (Figure 3). Therefore, it is most likely that in the flagship centers, PϽ.05) in lieu of extremity do- this robust difference is the product of one or more very nor sites (97% vs 48% in the flagship centers, PϽ.05) if powerful (many-fold stronger than health status) un- possible (Table 2). This difference reflects my experi- measured confounder(s). An obvious candidate, by vir- ence that torso sites tend to be closer to the ablative site, tue of most MAHCs’ public mission of “caring for those require some repositioning or are less readily exposed, who remain uninsured,”1 might be a greater percentage are inconsistently bulky, have shorter and deeper pedicles, of poor patients. However, like chronologic status, in- and inflict greater donor morbidity than extremity sites dicators of socioeconomic status have consistently been like the radial forearm and fibula. Our embrace of the adversely associated with health status and hence in large anterolateral thigh donor site for extensive soft tissue de- part only dependently (indirectly) with poorer surgical fects (19% vs 0% in the flagship centers, PϽ.05), in lieu outcomes. Independent risk factors are by definition more of the more conventional rectus abdominis and latissi- powerful than the risk factors dependent on them. For mus dorsi donor sites (1% vs 21% in the flagship cen- this reason, chronologic and socioeconomic status are gen- ters, PϽ.05), is emblematic of this approach. This ex- erally less powerful than health status, the usual under- tremely versatile, extensive soft tissue donor site is farther lying independent risk factor. Another obvious candi- from the ablative site, requires no repositioning, is readily date might be resident teaching status.1 However, we exposed, is moderately and uniformly thick, has a long currently sponsor 16 accredited residency programs, in- superficial pedicle, is simple to harvest (with a cuff of me- cluding medicine, surgery, obstetrics/gynecology, pedi- dial vastus lateralis), and inflicts minimal donor mor- atrics, and family practice (167 postgraduates). This is bidity.31 Although our preferred workhorse donor sites greater than the minimum to be considered an MAHC may be significantly different, we concur with an evolv- (4)1 and well within the flagship range (8-74).32 ing consensus at many of the flagship centers that when I believe, based on my experience (as a MRHNS fel- it comes to the total number in one’s armamentarium less low at a flagship center), that the most important un- may be more.4-7 measured confounder is a relative disconnection of financial risk between admitting physicians and their COMMENT hospitals. This disconnection is related to the fact that most HMOs (health insurance plans) today are both non- The objective of this study was to compare our outcomes integrated and for profit (plan accepts risks and rewards).2 with some of the best available. The implicit assumption Therefore, physicians and hospitals are insulated from was that high-experience and high-volume centers would each other’s risk by the virtual cornucopia of interven- have some of the best outcomes.3-7 The fact that all 5 flag- ing HMOs that insure their patients. The very nature of ship centers turned out to be highly ranked MAHCs1,22,23 our nonprofit, group-model, integrated comprehensive suggests that the selection methods were sound. Most health care delivery system2 means that all provider (phy- knowledgeable microvascular surgeons would agree that sician and hospital) costs must ultimately come from the (alphabetically) the M. D. Anderson Cancer Center in same pot (members’ prepaid premiums). Therefore, we Houston, Tex, Memorial Sloan-Kettering Cancer Center all have a direct stake in the financial fitness of our hos- and Mount Sinai Medical Center in New York, NY, Uni- pitals (and yes, even our health plan). Make no mistake. versity of California Medical Center in Los Angeles, and We are physicians first, and therefore achieving high qual- University of Pittsburgh Medical Center, Pittsburgh, Pa, ity for our patients is always the first and most impor- have some of the best MRHNS outcomes in the United tant consideration. However, as vested participants, States. However, most would also agree that integrated achieving high value (quality over cost) for our entire or-

(REPRINTED) ARCH OTOLARYNGOL HEAD NECK SURG/ VOL 130, JUNE 2004 WWW.ARCHOTO.COM 784

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 ganization is also an important consideration. It is for this neck reconstruction: a rational approach to donor site selection. Ann Plast Surg. reason that every MRHNS patient was evaluated every 2001;47:385-389. 8. Carey RG, Lloyd RC. Measuring Quality Improvement in Healthcare: A Guide to Sta- workday by their admitting HNSO, who was personally tistical Process Control Applications. New York, NY: Quality Resources; 1995:194. 11,16,24 motivated to achieve safe early discharge. It is this 9. Blackwell KE, Brown MT, Gonzalez D. Overcoming the learning curve in micro- same motivation that has led to implementation of sev- vascular head and neck reconstruction. Arch Otolaryngol Head Neck Surg. 1997; eral well-established efficiency measures (see the “Meth- 123:1332-1335. ods” section) and perhaps most importantly a target 6-day 10. Epstein AM. Volume and outcome—it is time to move ahead [editorial]. N Engl 24 J Med. 2002;346:1161-1164. inpatient CCP. 11. Ryan MW, Hochman M. Length of stay after free flap reconstruction of the head I remind the reader that LOS was only a selected in- and neck. Laryngoscope. 2000;110(2, pt 1):210-216. dicator of cost. Although we believe our approach in- 12. Urken ML, Buchbinder D, Weinberg H, et al. Functional evaluation following mi- creases value, since hospitalization remains the “big ticket” crovascular oromandibular reconstruction of the oral cancer patient. Laryngo- item in health care,16 this may not necessarily be true. scope. 1991;101:935-950. 13. Kroll SS, Schusterman MA, Reece GP. Costs and complications in mandibular Prospectively controlled, longitudinal studies that use reconstruction. Ann Plast Surg. 1992;29:341-347. more direct measures of global costs would be required 14. Hidalgo DA, Rekow A. A review of 60 consecutive fibula free flap mandible re- (although probably impractical) to definitively address constructions. Plast Reconstr Surg. 1995;96:585-596. the issue of cost across the integrated HMO and MAHC 15. Blackwell KE, Azizzadeh B, Ayala C, Rawnsley JD. Octogenarian free flap reconstruction: complications and cost of therapy. Otolaryngol Head Neck Surg. 2002; settings. However, this study at least suggests one im- 126:301-306. portant benefit of integration (physicians and hospitals 16. Robinson JC. Decline in hospital utilization and cost inflation under managed care sharing financial risk). We are able to return our pa- in California. JAMA. 1996;276:1060-1064. tients safely home to their families sooner. 17. Hahn GJ, Meeker WQ. Statistical intervals for proportions and percentages (bi- In conclusion, for high-experience and high- nomial distribution). In: Hahn GJ, Meeker WQ, eds. Statistical Intervals: A Guide for Practitioners. New York, NY: Wiley & Sons; 1991:104-105. volume centers, patients undergoing MRHNS at a large 18. Armitage P, Berry G, Mathews JS. Comparison of several groups. In: Armitage integrated HMO can expect morbidity and mortality out- P, Berry G, Mathews JS, eds. Statistical Methods Medical Research. 4th ed. Ox- comes similar to flagship centers in the United States, with ford, England: Blackwell; 2002:234-235. shorter hospitalizations. 19. Kroll SS, Evans GR, Goldberg D, et al. A comparison of resource costs for head and neck reconstruction with free and pectoralis major flaps. Plast Reconstr Surg. 1997;99:1282-1286. Submitted for publication February 17, 2003; final revi- 20. Tandberg D. Improved confidence intervals for the difference between two pro- sion received October 27, 2003; accepted November 25, 2003. portions. Centre for Evidence Based Medicine. Available at: http://www.cebm I thank Raoul Burchette, MS (senior data consultant, Re- .net/downloads.asp. Accessed October 18, 2003. search and Evaluation, Kaiser-Permanente), for his statisti- 21. Browne RH. Using the sample range as a basis for calculating sample size in power calculations. Am Stat. 2001;55:293-298. cal advice and guidance throughout this project. I also thank 22. The 2003 US News and World Report ranking of America’s best hospitals. Avail- Michael P. McNicoll, MD (senior HNSO), Frederic R. DiTirro, able at: http://www.usnews.com/usnews/nycu/health/hosptl/tophosp.htm. Ac- MD (chief of head and neck surgery), and Robert A. Kagan, cessed August 4, 2003. MD (chief of radiation oncology), for their support. 23. Listing of United States National Cancer Institute designated Comprehensive Can- Corresponding author and reprints: Edgar A. Lueg, MD, cer Centers. Available at: http://www3.cancer.gov/cancercenters/centerslist .html. Accessed August 4, 2003. FRCSC, 4900 Sunset Blvd, Suite 6C, Los Angeles, CA 90027 24. Lueg EA. Reducing hospitalization safely for microvascular surgery. Paper pre- (e-mail: [email protected]). sented at: Annual Meeting of the American Head and Neck Society; May 4, 2003; Nashville, Tenn. 25. Beausang ES, Ang EE, Lipa JE, et al. Microvascular free tissue transfer in elderly REFERENCES patients: the Toronto experience. Head Neck. 2003;25:549-553. 26. Piccirillo JF, Lacy PD, Basu A, Spitznagel EL. Development of a new head and 1. Association of American Medical Colleges Web site. Council of Teaching Hos- neck cancer-specific comorbidity index. Arch Otolaryngol Head Neck Surg. 2002; pitals and Health Systems (COTH). Available at: http://www.aamc.org/members 128:1172-1179. /coth/start.htm. Accessed August 4, 2003. 27. Mulholland S, Boyd JB, McCabe S, et al. Recipient vessels in head and neck mi- 2. Weiner JP, de Lissovoy G. Razing a tower of Babel: a taxonomy for managed crosurgery: radiation effect and vessel access. Plast Reconstr Surg. 1993;92: care and health insurance plans. J Health Polit Policy Law. 1993;18:75-103. 628-632. 3. Urken ML, Weinberg H, Buchbinder D, et al. Microvascular free flaps in head and 28. Bengtson BP, Schusterman MA, Baldwin BJ, et al. Influence of prior radio- neck reconstruction: report of 200 cases. Arch Otolaryngol Head Neck Surg. 1994; therapy on the development of postoperative complications and success of free 120:633-640. tissue transfers in head and neck cancer reconstruction. Am J Surg. 1993;166: 4. Schusterman MA, Miller MJ, Reece GP, Kroll SS, Marchi M, Goepfert H. A single 326-330. center’s experience with 308 free flaps for repair of head & neck cancer defects. 29. Kroll SS, Schusterman MA, Reece GP, et al. Choice of flap and incidence of free Plast Reconstr Surg. 1994;93:472-478. flap success. Plast Reconstr Surg. 1996;98:459-463. 5. Jones NF, Johnson JT, Shestak KC, Myers EN, Swartz WM. Microsurgical re- 30. Kroll SS, Robb GL, Reece GP, et al. Does prior irradiation increase the risk of construction of the head and neck: interdisciplinary collaboration between head total or partial free-flap loss? J Reconstr Microsurg. 1998;14:263-268. and neck surgeons and plastic surgeons in 305 cases. Ann Plast Surg. 1996;36: 31. Lueg EA. The anterolateral thigh flap: radial forearm’s “big brother” for exten- 37-43. sive soft-tissue head and neck defects. Arch Otolaryngol Head Neck Surg. In press. 6. Blackwell KE. Unsurpassed reliability of free flaps for head and neck reconstruc- 32. Accredited Residency Programs Kaiser Los Angeles Medical Center. Accredita- tion. Arch Otolaryngol Head Neck Surg. 1999;125:295-299. tion Council for Graduate Medical Education. Available at: http://www.acgme.org 7. Disa JJ, Pusic AL, Hidalgo DH, Cordeiro PG. Simplifying microvascular head and /adspublic. Accessed August 4, 2003.

(REPRINTED) ARCH OTOLARYNGOL HEAD NECK SURG/ VOL 130, JUNE 2004 WWW.ARCHOTO.COM 785