Assessing the relationships between health-related ...

ORIGINAL ARTICLE

European Journal of Cardio-Thoracic Surgery 44 (2013) 525–533 doi:10.1093/ejcts/ezt064 Advance Access publication 21 March 2013

Assessing the relationships between health-related quality of life and postoperative length of hospital stay after oesophagectomy for cancer of the oesophagus and the gastro-oesophageal junction† Philippe Nafteux*, Joke Durnez, Johnny Moons, Willy Coosemans, Georges Decker, Toni Lerut, Hans Van Veer and Paul De Leyn Department of Thoracic Surgery, University Hospitals Leuven, Leuven, Belgium * Corresponding author. Department of Thoracic Surgery, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium. Tel: +32-16346820; fax: +32-16346821; e-mail: [email protected] (P. Nafteux). Received 15 October 2012; received in revised form 20 December 2012; accepted 3 January 2013

Abstract OBJECTIVES: To evaluate baseline health-related quality of life (HRQL) factors that influence short-term outcome after oesophagectomy for cancer of the oesophagus and gastro-oesophageal junction and the effects of postoperative length of hospital stay on postoperative HRQL, as perceived by the patients themselves.

RESULTS: There were 372 males and 83 females, with a mean age of 63.1 years. Hospital mortality was 3.7% (17 patients). When analysing postoperative length of stay (LOS), a median of 10 days was found. In a multivariable analysis, using a binary logistic regression model, independent prognosticators for a longer LOS (>10 days) were: medical [hazard ratio, HR, 6.2 (3.62–10.56); P < 0.0001] and surgical [HR 2.79 (1.70–4.59); P < 0.0001] morbidity, readmittance to intensive care unit [HR 33.82 (4.55–251.21); P = 0.001] and poor physical functioning [HR 1.89 (1.14–3.14); P = 0.014]. Postoperatively, patients with early discharge (LOS 10 days. Return to the level of the reference population scores was achieved at 1 year in the LOS ≤10 days for almost all the scales, but not in the LOS >10 days group. CONCLUSIONS: A better perception of preoperative physical functioning might have a beneficial effect on LOS. Our data, furthermore, suggest that early discharge correlates with improved postoperative HRQL outcomes. A clear decrease of the HRQL is seen at 3 months after the surgery, particularly in the LOS >10 days group. Generally, return to the level of the reference population scores is achieved at 1 year in the LOS ≤10 days, but not in the LOS >10 days group. Keywords: Oesophageal neoplasms • Oesophagectomy • Quality of life • Length of stay

INTRODUCTION For oesophageal cancer, surgery remains the mainstay of treatment. Undoubtedly, oesophagectomy is a major surgical procedure, with high postoperative morbidity and mortality rates, with an average length of stay (LOS) in most series ranging from 11 to 26 days (mean 15) [1]. Although postoperative mortality has decreased substantially in high-volume centres, the risks of perioperative complications with the subsequent prolonged LOS remain high. Over the last years, there has been a growing interest in defining the risk factors for complications and prolonged †

Presented at the 26th Annual Meeting of the European Association for Cardio-Thoracic Surgery, Barcelona, Spain, 27–31 October 2012.

LOS; what these factors are and to what extent they can be influenced before and after oesophagectomy are still matters of debate. In recent decades, along with the increased success of oesophageal cancer treatment, health-related quality of life (HRQL) has become an increasingly considered outcome in cancer surgery in general and oesophageal cancer surgery in particular [2]. Although short-term HRQL is known to be negatively influenced by perioperative complications [3–5], the importance of preoperative HRQL as a predictor for postoperative evolution (including postoperative complications, LOS and postoperative HRQL) has not yet been fully evaluated. And inversely, the impact of prolonged LOS on the subsequent postoperative HRQL is still to be examined.

© The Author 2013. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.

THORACIC

METHODS: Four hundred and fifty-five patients operated on with curative intent between January 2005 and December 2009 were analysed. HRQL scores were obtained by European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (QLQ)-C30 and oesophageal-specific symptoms (OES-18) questionnaires at baseline (=day before surgery) and 3-monthly post-surgery for the first year.

526

P. Nafteux et al. / European Journal of Cardio-Thoracic Surgery

The aims of this study were to (i) identify which preoperative HRQL factors were associated with prolonged LOS following oesophagectomy for cancer of the oesophagus and gastro-oesophageal junction (GEJ) and (ii) investigate whether a shorter LOS correlates with improved HRQL outcomes over time postoperatively.

PATIENTS AND METHODS Study design All patients with a newly diagnosed oesophageal or GEJ cancer, operated on between January 2005 and December 2009 at the University Hospitals of Leuven, who underwent oesophagectomy with curative intent, were prospectively included. Exclusion criteria were the presence of R2 resection or metastasis at surgery and refusal or inability to fill in the HRQL questionnaires. LOS was measured from the day after surgery till the day of discharge. The median LOS after oesophagectomy in our department is 10 days. Therefore, we used this cut-off value to define normal (≤ 10 days) and prolonged LOS (> 10 days). The patients were followed up until death or the end of the study period (May 2012). The Ethics Committee at the University Hospitals of Leuven approved the study (S54541; ML8493).

Health-related quality-of-life data Several reference populations for HRQL scores in several European countries have been published, but a reference population for Belgium is lacking. As most of these reference population evaluations have been conducted >10 years ago, we decided to use the recently published references values from a random sample of the Swedish population [6]. HRQL scores were obtained by European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (QLQ)-C30 [7] and oesophageal-specific symptoms (OES-18) [8]. The HRQL questionnaires were filled in prospectively and independently by the patients (self-reporting) the day before surgery (baseline) and 3-monthly post-surgery for the first year. The QLQ-C30 [7] incorporates nine multi-item scales: one global quality-of-life (QoL) scale; five functional scales addressing physical capacity, role functioning at work and during leisure time, cognitive function, emotional functioning and social functioning; and three symptom scales regarding fatigue, pain and nausea. Six single-item symptom measures are also included (dyspnoea, insomnia, appetite loss, constipation, diarrhoea and financial problems). The OES-18 [9] assesses symptoms specific to oesophageal cancer. These symptoms include dysphagia, problems with swallowing and eating, and indigestion. The EORTC QLQ-C30 and OES-18 questionnaire responses were linearly transformed to scores from 0 to 100 in accordance with the EORTC Scoring Manual [10]. High scores in the function scales and the global QoL scale indicate better levels of function and QoL, respectively, whereas high scores in the symptom scales and items represent more severe symptoms. A difference of 10 points between the two groups was considered to be ‘clinically significant’. These criteria have been described previously [8] and have been used in multiple previous reports. All HRQL scales had four response categories: [1] ‘not at all’, [2] ‘a little’, [3] ‘quite a bit’ and [4] ‘very much’, except for the global QoL scale, which has a seven-step scale ranging from ‘very poor’ to ‘excellent’.

The responses to questions within each function scale were dichotomized into ‘good’ vs ‘poor’ function. As every question has four response categories, patients who responded 3 ‘quite a bit’ or 4 ‘very much’ to any question within a scale were categorized as having ‘poor’ function, while the others were categorized as having ‘good’ function for that scale (all response categories being 1 or 2 to every question within a scale). The responses to the questions forming the symptoms scales were also dichotomized into ‘no or minor symptoms’ vs ‘symptomatic’. Patients who had at least one response of 3 ‘quite a bit’ or 4 ‘very much’ for any question within a symptom scale and for a single item were categorized as being symptomatic for that symptom scale or item; other patients were categorized as having minor symptoms (all response to every question being 1 or 2 within a scale). On the global QoL scale (consisting of seven steps from very poor [1] to excellent [7]), a response of 4 or less (i.e. worse) to either of the two questions included in the scale was considered to represent a poor global QoL. Otherwise, the patient was categorized as having a good global QoL (answer to the two questions being 5 or more). This strategy was used to facilitate interpretation of the data and has been used in previous research [5, 7]. Patients with missing data for a particular scale were excluded from analyses of that individual scale, but were included in other scales and item analyses if the relevant data were available. Adjustments of the reference population values for age and gender were performed to match our studied population [11]. Since there were no such age categories as [30–39] and [80–89] specified in the reference population ( population ranging from 40 to 80 years of age) [9], three patients aged 80 (mean 82.8 years) were classified in the [70–79] group.

Statistical analyses Patient-related, surgical and oncological factors were compared between the two groups using either the Student’s t-test or χ 2 test where appropriate. Preoperative HRQL scores were compared with an age- and gender-adjusted reference population. A mean difference of 10 points was considered clinically significant [8]. Binary logistic regression models were built to determine independent prognostic factors regarding prolonged LOS and postoperative poor physical functioning. Hazard ratio (HR), 95% confidence interval for HR and significance were reported. The Hosmer and Lemeshow test was used to determine goodness-of-fit of the models; χ² (df) and significance are reported for each model. The significance level was set at 0.05. For all data analyses, the statistical software packages IBM SPSS Statistics 19 for Windows and JMP 8 (SAS Institute 2009) were used.

RESULTS In our prospective database, we reviewed 490 consecutive patients meeting the inclusion criteria. Thirty-five of these were excluded (Table 1). Four hundred and fifty-five patients were included for analysis of the perioperative data and preoperative HRQL questionnaires. There were 372 males and 83 females, with a mean age of 63.1 (range 34–88 years). One hundred and thirty-three (29%) patients underwent induction chemoradiation


January 2005–December 2009 490 consecutive patients underwent oesophagectomy - 35 no preoperative HRQL questionnaire (not interested, mental status, …) Analysis preoperative HRQL data 455 (93%) patients completed preoperative HRQL questionnaire Analysis at 1-year postoperative - 68 died before 12 months - 13 too ill/hospitalized at 12 months postoperatively - 44 did not return HRQL questionnaire 330 (73%) patients completed both pre- and postoperative HRQL questionnaires Bold is used to highlight the number of patients involved.

therapy prior to surgery, and the remaining 322 (71%) patients underwent primary surgery. Adenocarcinoma was found in 324 (71%) patients. R0 resection was achieved in 429 (94%) patients, while 26 (6%) underwent R1 resection. Union Internationale Contre le Cancer (UICC) pTNM seventh edition distribution was 14.5, 35.2, 17.4, 30.3 and 2.6% for Stages 0, I, II, III and IV, respectively. Medical complications were found in 288 (63%) of the patients and consisted mostly of respiratory (164 or 36%) or cardiac (61 or 13%) complications. Surgical complications were seen in 171 (38%) patients. Most frequently observed were prolonged fluid drainage from thoracic drain (42 or 9%) and pleural effusions requiring drainage (39 or 9%). Cervical wound problems with suspicion of anastomotical leakage or real clinical anastomotical leakage were seen in 27 (or 6%). Major morbidity, requiring (re-) admittance to intensive care unit (ICU), was seen in 52 (11%) patients, mostly due to major respiratory complications (47 of 52 or 90%). Hospital mortality was 3.7% (17 patients), including 1 patient who died 13 months after surgery, due to multiple respiratory complications. Overall survivals were 84.4 and 50.2% at 12 and 60 months, respectively. Disease-free survival was 44.5% at 5 years. There were 231 (51%) patients in the group LOS ≤10 days and 224 (49%) in the group >10 days. A significant (P = 0.0047) survival benefit was noted in patients with an LOS of ≤10 days (57.1 vs 41.4% at 60 months) unexplainable by differences in pathological staging.

Effect of perioperative data and preoperative health-related quality-of-life questionnaires on length of stay On univariable analysis, there were no significant differences found in patient-related [age, sex, preoperative BMI, preoperative weight loss and American Society of Anesthesiologists (ASA) score], surgical (type of surgery, blood loss, duration of surgery and R-status) or oncological (pStage, lymph-node status, number of resected/positive lymph nodes and grading) factors between both groups (LOS ≤10 vs >10 days) (Table 2). Analyses of the mean preoperative HRQL scores (Table 3) revealed a clinically significant decrease (> 10 points) in global HRQL, emotional function and increase in appetite loss in the studied population compared with the age- and gender-adjusted reference population. Moreover, in patients with LOS >10 days, there was also a clinically significant decrease in role functioning.

As a refinement of the mean HRQL scores, all mean scores were further dichotomized in good/poor functioning or the absence/presence of symptoms (Table 4). When analysing the percentage of patients with poor function or presenting with symptoms, there were more patients in the study population with poor function/symptoms compared with the reference population. In all patients, a higher percentage of patients showed poor global HRQL and emotional function together with an increased appetite loss. This is even more prominent in the group of patients with LOS >10 days, where additionally, a poor physical function was observed. Multivariable analysis, using a binary logistic regression model, showed significant differences for medical [HR 6.2 (3.62–10.56); P < 0.0001] and surgical [HR 2.79 (1.70–4.59); P < 0.0001] morbidity, readmittance to ICU [HR 33.82 (4.55–251.21); P = 0.001] and poor physical functioning [HR 1.89 (1.14–3.14); P = 0.014], being more frequently observed in the LOS >10 days group. These four variables are independent prognosticators for a longer LOS. Not withheld in the multivariate models were: induction chemoradiotherapy (P = 0.10), poor global QoL (P = 0.13), poor social functioning (P = 0.27), poor role functioning (P = 0.60), poor emotional functioning (P = 0.56) and poor cognitive functioning (P = 0.37). The Hosmer and Lemeshow test (test of goodness-of-fit) shows that model fit is acceptable χ² [7] = 12.23, P = 0.093.

Effects of length of stay on postoperative health-related quality of life When compared with preoperative health-related quality of life. Three hundred and thirty patients who completed both preoperative and 1-year postoperative questionnaires were retained for the analysis of postoperative HRQL. Patients who refused to participate to the study were not alive at 1 year or failed to fill in preoperative of 1 year questionnaire were excluded from the analysis. In general, at 1-year post-oesophagectomy, baseline levels were not regained for physical and role functioning and several symptom scales [fatigue, dyspnoea, diarrhoea, financial difficulties, gastrointestinal (GI) symptoms and couching] remained clinically significantly lower compared with preoperative scales in the LOS ≤10 days group. Furthermore, patients with a LOS of >10 days showed additional lower scores for cognitive functioning and higher scores for pain, nausea/vomiting, eating problems and speech problems.

When compared with the reference population. A clear impact of the treatment was seen in the functional and symptom scales for all groups at 3 months (Fig. 1). This decrease was generally less marked in the LOS ≤10 days group. In this group, a clinically significant decrease (>10 points difference) was seen at the level of physical functioning, role functioning, cognitive functioning, social functioning, fatigue, nausea/vomiting, dysphagia and dry mouth. In the LOS>10 days group, the negative impact reached the clinically significant level in almost all functional and symptoms scales. In the LOS10 days (N = 224)

63.12

[34–88]

62.68

[34–88]

63.57

[37–88]

372 83 25.48 6.3 133

82% 18% [15, 1–44, 2] [−21–70] 29%

197 34 25.12 5.9 56

85% 15% [15, 9–39, 6] [−6–70] 24%

175 49 25.84 6.7 77

78% 22% [15, 1–44, 2] [−21–50] 34%

63 305 87

14% 67% 19%

40 154 37

17% 67% 16%

24 155 51

11% 67% 22%

333 29 21 72 431 333

73% 6% 5% 16% [40–1950] [180–650]

172 11 14 34 404 328

74% 5% 6% 15% [85–1300] [180–650]

161 18 7 38 458 338

72% 8% 3% 17% [40–1950] [180–490]

429 26

94% 6%

218 13

94% 6%

211 13

94% 6%

66 160 79 138 12

15% 35% 17% 30% 3%

28 81 43 72 7

12% 35% 19% 31% 3%

38 79 36 66 5

17% 35% 16% 29% 2%

271 184 23.73 2.06

60% 40% [1–85] [0–44]

129 102 23.15 2.09

56% 44% [1–80] [0–44]

142 82 24.33 2.04

63% 37% [1–85] [0–44]

63 112 183 2 94 288 164 61 171 27 52

14% 25% 40% 0% 21% 63% 36% 13% 38% 6% 11%

32 63 96 0 40 90 44 22 45 1 1

14% 27% 42% 0% 17% 39% 19% 10% 19% 0% 0%

31 49 87 2 54 198 120 39 126 26 51

14% 22% 39% 1% 24% 88% 54% 17% 56% 12% 23%

P-value 0.36 0.056

0.10 0.29 0.018 0.87

0.70

0.073 0.13 0.94

0.21

0.10

0.34 0.90 0.07

10 (N = 224)

Obs

Ref°

Diff

Obs

Ref

Diff

67.6 89.6 84.3 75.8 89.4 87.4 20.7 7.9 12.5 12.7 21.6 15.6 11.7 4.7 7.4 21.2 9.5 21.5 9.1 14.0 14.4 9.0 14.6 3.9

77.0 89.7 89.1 86.2 88.3 91.6 18.0 2.5 17.5 16.1 16.2 3.3 4.4 5.6 4.6 na na na na na na na na na

−9.4 0.0 −4.7 −10.4 a 1.1 −4.2 2.7 5.5 −5.0 −3.4 5.4 12.3 a 7.3 −0.9 2.8

63.9 83.3 78.2 74.0 86.9 83.4 25.5 10.2 13.2 16.8 24.7 17.4 11.5 6.5 9.7 23.0 9.6 22.8 10.7 14.9 20.4 13.5 21.1 4.9

76.8 89.3 88.8 86.0 88.2 91.4 18.3 2.5 17.9 16.1 16.7 3.3 4.8 5.6 4.6 na na na na na na na na na

−12.9 a −5.9 −10.6 a −11.9 a −1.3 −8.0 7.2 7.7 −4.7 0.7 8.0 14.1 a 6.7 0.9 5.1

a

THORACIC

10 points difference. Bold and italic values indicate clinically significant difference ≥10%. QoL: quality of life; GI: gastrointestinal; obs: observed; ref: reference population; diff: difference; na: not available.

Table 4: Comparing the percentage of patients indicating a preoperative poor functioning or being symptomatic to an age- and gender-adjusted reference population % of patients indicating preoperative poor function or being symptomatic compared with the age-/gender-adjusted reference population All (N = 455)

Poor global QoL Poor physical function Poor role function Poor emotional function Poor cognitive function Poor social function Symptomatic fatigue Symptomatic nausea/vomitting Symptomatic pain Symptomatic dyspnoea Symptomatic insomnia Symptomatic appetite loss Symptomatic constipation Symptomatic diarrhoea Symptomatic financial difficulties a

LOS ≤10 (N = 231)

LOS >10 (N = 224)

Obs (%)

Ref (%)

Diff (%)

Obs (%)

Ref (%)

Diff (%)

Obs (%)

Ref (%)

Diff (%)

30.3 29.0 16.9 36.9 11.6 14.3 25.7 9.7 13.0 8.4 18.5 13.8 8.8 2.6 11.6

16.9 19.6 12.1 18.4 10.6 8.1 22.0 2.3 19.1 8.9 13.6 2.0 2.9 3.2 4.6

13a 9 5 18a 1 6 4 7 −6 −1 5 12a 6 −1 7

24.2 22.1 14.3 34.2 9.5 11.3 22.1 9.1 12.6 5.2 17.3 11.3 8.2 2.2 9.5

16.9 19.8 12.2 18.6 10.7 8.1 22.1 2.3 19.3 8.9 13.7 2.1 2.9 3.2 4.6

7 2 2 16a −1 3 0 7 −7 −4 4 9 5 −1 5

36.6 36.2 19.6 39.7 13.8 17.4 29.5 10.3 13.4 11.6 19.6 16.5 9.4 3.1 13.8

16.8 19.5 12.1 18.3 10.6 8.1 21.8 2.3 18.9 8.8 13.5 2.0 3.0 3.2 4.6

20a 17a 8 21a 3 9 8 8 −6 3 6 14a 6 0 9

10% difference. Bold and italic values indicate clinically significant difference ≥10%. QoL: quality of life; LOS: length of stay; obs: observed; ref: reference population; diff: difference.

530


Figure 1: Comparison of the evolution of global QoL, physical functioning, role functioning and fatigue for the groups LOS≤10 days and LOS >10 days. LOS: length of stay; preop: preoperative; postop: postoperative; yr: year.

technique, this relationship still existed; patients with good function/symptom had more chance of presenting better function and fewer symptoms at 1 year, and the opposite was true for patients presenting poor preoperative function/symptom scores. On the other hand, still a non-unimportant group of patients are so negatively influenced by the treatment that a change from good function/symptom state towards the group poor function/symptom state is noted. However, since preoperative poor physical functioning was an independent prognosticator for LOS, we decided to examine more closely physical functioning (Table 6). In the study population, the majority of patients presented a good preoperative physical functioning (n = 241 or 73%). 64% of those patients retained a good physical functioning at 1 year (n = 155 or 47% of the total population). On the other hand, a minority of patients had a poor preoperative physical functioning (n = 89 or 27%). Of those, 19 patients improved their physical functioning enough to be categorized as good physical functioning (6% of total population). We performed a multivariable analysis to determine which factor could play a role in the development of poor physical functioning. Readmission to the ICU [HR 5.19 (1.94–13.89); P = 0.001], preoperative poor physical [HR 5.94 (3.26–10.82); P < 0.0001] function and recurrent disease [HR 2.81 (1.41–5.60); P = 0.003] were independent prognosticators for poor 1-year postoperative physical functioning (Hosmer and Lemeshow test χ² [4] = 1.653, P = 0.8). The LOS does not seem to be an independent prognosticator (P = 0.409; induction chemoradiation therapy P = 0.44).

DISCUSSION Oesophageal cancer is an increasingly common cancer and, in particular for adenocarcinoma, is now the fastest-rising solid tumour in most Western countries. Surgery-based treatment remains the treatment of choice for potentially curable oesophageal cancers. Undoubtedly, oesophagectomy is a major surgical procedure with high morbidity and often a long LOS. Despite improvements in the last decades, the postoperative period remains often complicated, resulting in prolonged LOS. Those factors combined with survival represented the main (and often only) outcome measures that could be used to evaluate the treatments for oesophageal cancer. However, in recent years, along with increased success of therapies, HRQL has generally become accepted as another important outcome parameter. Although an increased number of series on HRQL has been published, the importance of preoperative HRQL for the subsequent postoperative evolution and, in particular, its impact on LOS and postoperative HRQL have not yet been fully evaluated. In this prospective study, we were interested in the relationship between preoperative HRQL and the LOS post-surgery and the relationship between that LOS and the postoperative HRQL evolution at 1 year post-surgery. As about 15% of patients did not survive long enough to reach 1 year, we reviewed the data of this group of patients (data not shown) looking for differences with the group of patients surviving at least 1 year. Preoperative functional scales


531

Table 5: Postoperative findings for groups LOS >10 days vs LOS ≤10 days Comparing HRQL data preop vs 1-year postop

QoL Physical functioning Role functioning Emotional functioning Cognitive functioning Social functioning Fatigue Nausea/vomiting Pain Dyspnoea Insomnia Appetite loss Constipation Diarrhoea Financial difficulties Dysphagia Deglutition Eating GI symptoms GI pain Dry mouth Taste Couching Speech problems

LOS ≤10 days

LOS >10 days

Preop (N = 330)

Postop 1 yr (N = 330)

Diff

Preop (N = 180)


Diff

Preop (N = 150)


Diff

71.4 87.5 82.8 74.7 89.1 85.9 21.5 7.7 11.9 13.2 22.7 13.5 11.0 5.3 7.4 21.5 8.8 20.8 10.4 13.5 16.9 10.7 16.1 4.5

64.0 76.0 69.2 75.3 81.5 77.6 37.5 17.2 23.0 25.7 26.4 22.7 11.3 19.7 19.1 20.8 15.8 31.7 26.4 14.8 23.4 17.7 27.3 11.8

−7 −11 a a −14 1 −8 −8 16 a 10 11 a 12 a 4 9 0 14 a 12 a −1 7 11 a 16 a 1 7 7 11 a 7

72.7 90.3 85.9 76.0 90.3 88.1 18.5 7.0 11.4 10.7 21.3 12.0 11.1 4.1 6.1 20.6 8.6 20.8 9.9 13.6 14.1 8.0 12.6 4.1

67.1 79.8 73.7 77.4 85.0 80.6 34.8 15.7 20.2 21.4 24.2 21.7 9.1 19.9 18.0 18.4 14.9 30.8 24.9 14.1 19.9 16.0 25.8 8.1

−6 −10 a −12 a 1 −5 −7 16 a 9 9 11 a 3 10 −2 16 a 12 a −2 6 10 15 a 1 6 8 13 a 4

69.8 84.1 79.0 73.1 87.7 83.2 25.1 8.4 12.6 16.2 24.4 15.3 10.9 6.7 8.9 22.7 9.0 20.8 10.9 13.4 20.2 14.0 20.2 5.1

60.9 71.4 63.8 72.7 77.2 73.9 40.8 19.0 26.3 31.0 29.0 23.9 14.0 19.5 20.5 23.8 16.9 32.9 28.1 15.6 27.6 19.7 29.2 16.2

−9 −13 a −15 a 0 −10 a −9 16 a 11 a 14 a 15 a 5 9 3 13 a 12 a 1 8 12 a 17 a 2 7 6 9 11 a

a

10% difference. Bold and italic values indicate clinically significant difference ≥10%. QoL: quality of life; GI: gastrointestinal; HRQL: health-related quality of life; preop: preoperative; postop: postoperative; yr: year; LOS: length of stay; diff: difference.

Effect on length of stay of perioperative data and preoperative health-related quality of life questionnaires

Table 6: Evolution of physical functioning Preop good PF

One-yr postop good PF One-yr postop poor PF

Preop poor PF

n

%

n

%

155 86

47 26

19 70

6 21

PF: physical functioning; preop: preoperative; postop: postoperative; yr: year.

data are comparable between patients who died during, or survived, the first year. For the symptom scales, only fatigue, nausea/vomiting, pain and appetite loss seem to be more pronounced in the group who died during the first year. There was also no difference noted when dividing this group according to the LOS. Furthermore, the evolution of the HRQL in the group of patients who died during the first year follows the same trend at 3 and 6 months as the group of patients surviving the first year. On the contrary, the HRQL of the group of patients who did not survive at 1 year does not seem to recover after 3 or 6 months, but seems to decrease further as the general condition deteriorates, leading to death.

A clear impact of the disease on the whole group was seen at the level of the preoperative HRQL score when compared with the reference population. This was particularly marked for global HRQL, emotional function and appetite loss. Although the scores in the LOS >10 days group showed in general poorer results when compared with the LOS ≤10 days group, only role and physical functioning (after dichotomization) were significantly decreased when compared with the LOS ≤10 days group. Those differences could be due to a more pronounced impact of the disease or possibly the presence of more important co-morbidities in this group of patients. The impact of co-morbidities on the evolution of HRQL has been shown by others [12]. In the present study, we could not confirm such a relation. In the multivariable analysis, postoperative complications, readmittance to ICU and poor preoperative physical functioning were the only predictors for prolonged LOS. Although complications are well-known predictors of outcome after oesophagectomy [13], the importance of physical functioning should not be underestimated, as described previously [14, 15]. Therefore, pre- and postoperative revalidation programmes [16] and aggressive treatment of malnutrition and cachexia by means of supplementary feeding [17], particularly in the case of neoadjuvant treatment, could positively influence the physical functioning of the patient,

THORACIC

All

532


eventually shortening the LOS by limiting complications, enhancing postoperative HRQL and survival [15, 18].

Effects of length of stay on postoperative health-related quality of life In general, at 1-year post-oesophagectomy, patients did regain their baseline levels. In the LOS ≤10 days group, preoperative levels were not regained for physical and role functioning and several symptom scales (fatigue, diarrhoea, financial difficulties, GI symptoms and couching). Patients with a LOS of >10 days, however, showed additional lower scores for cognitive and social functioning and higher scores for pain, dyspnoea, eating problems and speech problems when compared with their preoperative scores. Important differences were noted when comparing both LOS groups in regard to the reference population. In general, in the LOS 10 days group were still negatively influenced at 1 year. Those differences persisted when we compared both LOS groups. So the negative impact of the treatment was less present when patients could be discharged early and they seemed to lead a satisfactory life despite the oesophageal cancer treatment. A larger impact was noted in the case of a prolonged LOS, and patients generally did not recover even after 1 year. This impact has been noted by others [19, 20]; we noted a more explicit impact during the first 6 months after the surgery followed by a recovery up to almost reference population but not completely. Explanations for those lower scores at 1 year when compared with the reference population have been proposed in the literature. One suggested factor is difference in age. As in our study the reference population was adjusted for age and gender (factors known to have an important impact on HRQL in a population [11]), the effect of this factor is, in this series, limited. Another suggested explanation could be the lower preoperative HRQL in the study group when compared with the reference population. In our study, it does not seem to be the case as the HRQL scores were in general comparable with the reference population since only 3 of 18 scales were clinically significantly different (global QoL, emotional functioning and appetite loss). We believe that the changes noted at the level of postoperative HRQL are true effects of the treatment, but the effective causes of those changes can be multifactorial.

population. On multivariable analysis, poor physical functioning, readmission to ICU (due to major complications) and recurrence were predictors of poor physical functioning at 1 year. This observation is confirmed by others [13, 19]. The results of this study emphasize the need to adopt therapeutic strategies to enhance the disease-free period and to avoid postoperative complications (e.g. by centralization of this delicate and high-risk surgery) [21, 22]. Furthermore, although the impact of the findings of this study on potential changes in treatment strategy is difficult to define, one can only postulate that including high-risk patients in preoperative rehabilitation programmes (e.g. respiratory rehabilitation, feeding support and psychological support) could be beneficial. If supported by further studies, this could be a powerful message for the Healthcare System Representatives and Hospital Directors. On the other hand, great care must be taken not to delay treatment unnecessarily, since that could compromise the oncological outcomes. The possible interaction between all those variables should be further evaluated before definitive conclusions can be drawn.

CONCLUSIONS Our data suggest that early discharge correlates with improved postoperative HRQL outcomes. A better perception of preoperative HRQL and global health status by patients themselves, together with a better preoperative role and social functioning, might have a beneficial effect on LOS. Poor preoperative physical functioning and postoperative complications (in particular readmission to ICU) each contributed to a prolonged LOS in a multivariable model. A clear decrease of the HRQL is seen at 3 months after the surgery and is clearly more pronounced in the LOS >10 days group. Generally, recovery of the preoperative score at 1 year is achieved in the LOS ≤10 days, but not in the LOS >10 days group. Preoperative physical functioning appears to be a major prognosticator, not only of the LOS, but also of postoperative physical functioning, besides re-admission to ICU and recurrence. Therefore, in order to improve postoperative HRQL, all efforts should be directed at increasing the perception of preoperative HRQL and further decreasing post-surgical morbidity, resulting in a decreased length of hospital stay. Conflict of interest: none declared.

REFERENCES Effects of preoperative health-related quality of life on postoperative health-related quality of life As physical functioning is the only HRQL parameter predictor for prolonged LOS in this study, we focussed more specifically on this functional scale. Physical functioning was not completely recovered at 1 year, particularly in the LOS >10 days group. Also, a clear relationship exists between good preoperative and postoperative physical functioning scores. Nevertheless, a large number of patients with preoperative good physical functioning were negatively influenced by the treatment and did fully recover even after 1 year. On the contrary, a small number of patients with poor physical functioning were positively influenced by the treatment and could improve their scores within range of the reference

[1] Metzger R, Bollschweiler E, Vallböhmer D, Maish M, DeMeester TR, Hölscher AH. High volume centers for esophagectomy: what are the number needed to achieve low postoperative mortality? Dis Esophagus 2004;17:310–4. [2] Parameswaran R, McNair A, Avery KN, Berrisford RG, Wajed SA, Sprangers MAG et al. The role of health-related quality of life outcomes in clinical decision making in surgery for esophageal cancer: a systematic review. Ann Surg Oncol 2008;15:2372–9. [3] de Boer AG, van Lanschot JJ, van Sandick JW, Hulscher JB, Stalmeier PF, de Haes JC et al. Quality of life after transhiatal compared with extended transthoracic resection for adenocarcinoma of the esophagus. J Clin Oncol 2004;22:4202–8. [4] Reynolds JV, McLaughlin R, Moore J, Rowley S, Ravi N, Byrne PJ. Prospective evaluation of quality of life in patients with localized oesophageal cancer treated by multimodality therapy or surgery alone. Br J Surg 2006;93:1084–90.

[5] Djärv T, Lagergren J, Blazeby JM, Lagergren P. Long term health-related quality of life following surgery for oesophageal cancer. Br J Surg 2008; 95:1121–6. [6] Derogar M, Van Der Schaaf M, Lagergren P. Reference values for the EORTC QLQ-C30 quality of life questionnaire in a random sample of the Swedish population. Acta Oncol 2012;51:10–6. [7] Blazeby JM, Alderson D, Winstone K, Steyn R, Hammerlid E, Arraras J et al. Development of an EORTC questionnaire module to be used in quality of life assessment for patients with oesophageal cancer. The EORTC Quality of Life Study Group. Eur J Cancer 1996;32:1912–7. [8] Osoba D, Rodrigues G, Myles J, Zee B, Pater J. Interpreting the significance of changes in health-related quality-of-life scores. J Clin Oncol 1998;16:139–44. [9] Blazeby JM, Conroy T, Hammerlid E, Fayers P, Sezer O, Koller M et al. European Organisation for Research and Treatement of Cancer Gastrointestinal and Quality of Life Groups. Clinical and psychometric validation of an EORTC questionnaire module, the EORTC QLQ-OES18, to assess quality of life in patients with oesophageal cancer. Eur J Cancer 2003;39:1384–94. [10] Aaronson NK, Ahmedzai S, Bergman B, Bullinger M, Cull A, Duez NJ et al. The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 1993;85:365–76. [11] Hjermstad MJ, Fayers PM, Bjordal K, Kaasa S. Using reference data on quality of life—the importance of adjusting for age and gender, exemplified by the EORTC QLQ-C30 (+3). Eur J Cancer 1998;34:1381–9. [12] Djärv T, Blazeby JM, Lagergren P. Predictors of postoperative quality of life after esophagectomy for cancer. J Clin Oncol 2009;27:1963–8. [13] Derogar M, Orsini N, Sadr-Azodi O, Lagergren P. Influence of major postoperative complications on health-related quality of life among long-term survivors of esophageal cancer surgery. J Clin Oncol 2012;30:1615–9. [14] Quinten C, Coens C, Mauer M, Comte S, Sprangers MA, Cleeland C et al. Baseline quality of life as a prognostic indicator of survival: a meta-analysis of individual patient data from EORTC clinical trials. Lancet Oncol 2009;10:865–71. [15] van Heijl M, Sprangers M, de Boer A, Lagarde S, Reitsma H, Busch O et al. Preoperative and early postoperative quality of life predict survival in potentially curable patients with esophageal cancer. Ann Surg Oncol 2010;17:23–30. [16] Lococo F, Cesario A, Sterzi S, Magrone G, Dall’armi V, Mattei F et al. Rationale and clinical benefit of an intensive long-term pulmonary rehabilitation program after oesophagectomy: preliminary report. Multidiscip Respir Med 2012;7:21. [17] Koretz R. Should patients with cancer be offered nutritional support: does the benefit outweight the burden? Eur J Gastroenterol Hepatol 2007;19:379–82. [18] Djärv T, Lagergren P. Six months postoperative quality of life predicts long-term survival after oesophageal cancer surgery. Eur J Cancer 2011; 47:530–5. [19] Scarpa M, Valente S, Alfieri R, Cagol M, Diamantis G, Ancona E et al. Systematic review of health-related quality of life after esophagectomy for esophageal cancer. World J Gastroenterol 2011;17:4660–74. [20] Courrech Staal E, van Sandick J, van Tinteren H, Cats A, Aaronson NK. Health-related quality of life in long-term esophageal cancer survivors after potentially curative treatment. J Thorac Cardiovasc Surg 2010;140: 777–83. [21] Rouvelas I, Lagergren J. The impact of volume on outcomes after oesophageal cancer surgery. ANZ J Surg 2010;80:634–41. [22] Dimick JB, Pronovost PJ, Cowan JA, Lipsett PA. Surgical volume and quality of care for esophageal resection: do high-volume hospitals have fewer complications? Ann Thorac Surg 2003;75:337–41.

APPENDIX. CONFERENCE DISCUSSION Dr D. Wood (Seattle, WA, USA): When I looked through your paper and saw your presentation, you have an enormous amount of very intriguing data on health-related quality of life and I can see where this is going. You have such a rich database now that I know we are going to see more studies of health-related quality of life in oesophageal cancer patients. I have not seen these kinds of numbers and this kind of robust database of quality of life in this patient population. I have a couple of observations and a question. Certainly the increased length of stay leads to a decreased quality of life, both short-term and longterm, but I think it is important to recognize that length of stay is just a

533

surrogate for a variety of poor outcomes, obviously. It is not the length of stay itself, meaning if you could throw them out of the hospital at day 9, it is not going to make them fine. The point is that it is a surrogate for everything else that is going on that is holding them in the hospital, ranging from frailty or poor function to complications. I am surprised that there is not more impact of the preoperative quality of life on the length of stay; I would have expected a more profound difference. I wonder whether you picked the right cut point. This is my biggest question, which is, you used your median length of stay as the cut point for short or long, yet there is not really a difference between a 9-day hospitalization and an 11-day hospitalization. There is a difference between a 9-day hospitalization and a 19-day hospitalization. The Society of Thoracic Surgeons identified 14 days as an outlier for lobectomy, as an example. What do you think the impact would be if you chose outliers for prolonged length of stay rather than simply cutting your patient population in half, with the half that was out early and the half that was out late? I think that you might not have detected the differences that you could have if you had dichotomized your patient population more aggressively. Dr Durnez: I think your comment is correct, but a cut-off is also an artificial point, and there is no generally-accepted definition of what represents a prolonged length of stay. There are great ranges in different publications from different centres. We just had to choose one point and we decided to choose a median length of stay, so we had two groups of similar numbers. But I do realize that it might be different if we took another cut-off. It is maybe something we can investigate in the future. Dr Wood: Yes, I might suggest trying that, and maybe you already have, because I think that you might find a more profound power of preoperative health-related quality of life in predicting postoperative outcomes. My only other question is, how do you feel that the non-responders at one year may have skewed your data; that is, were non-responders more likely to be poorly-functioning patients and therefore they didn’t respond, so you lost them at follow-up? Dr Durnez: It is correct that at one year, we only had 330 patients and we started with 455, so we lost 125, but we do know where they are: 68 of them died, 11 of them were hospitalized for chemotherapy, so they had recurrence at one year postoperatively, and 44 just chose not to answer. They came to the outpatient clinic for control. They were not having recurrence, they were not dead, but they just did not answer the questions. So it means that if those 68 patients who died would have died two weeks later when they would have come to the one-year control, they probably would have had poor physical function, and so that would make the impact of recurrence more profound than it is now, that is correct. Dr Wood: In some studies like this, there are correction factors for extrapolating non-responders and assuming that they are negative, and that would be another thing to consider in extrapolating a response for these nonresponders. I don’t know all the statistical details of that, but certainly that is done in other aspects of quality of life measurements. Dr Durnez: Yes. Dr K. Moghissi (East Yorkshire, UK): In my experience, there are two factors about which I would like your comment: One, is there a relationship between the stage of the tumour and what follows, and then, secondly, the aftercare from discharge to wherever. That is the issue at the moment in most of our countries. Once upon a time, you would have the patient in, say, for 12 days and send them to a convalescent home or something like that, but today the patient is discharged into the community, and the community, the organisation, is not sufficiently robust to care for these types of patients, and that, therefore, must undoubtedly affect their quality of life. What is your experience with that? Dr Durnez: Sorry, I did not really understand. Dr Moghissi: The question of aftercare, not within the hospital, but once discharged from the hospital, the majority of the patients suffer and the quality of life is affected because the care provided post-discharge from the hospital is not sufficient, and it changes from one region to another. Do you see what I mean? Toni, do you see? Did I explain it well? The patient stays 10 days or 12 days and goes home and you are measuring the quality of life, but it is the postoperative follow-up care within the community which will affect the actual quality of life of the patient, not the time that the patient stays at the hospital. Dr T. Lerut (Leuven, Belgium): It is a great remark, but it is unmeasurable at this point, and it is certainly also something which differs from country to country, from continent to continent. These are important things. This is also equally valid in the preoperative care of the patients, as Dr Durnez alluded to, and we now have oncological nurses and all these people who are coming into the system to improve the way that patients are taken care of both preoperatively and postoperatively, but to measure that at this point is extremely difficult.

THORACIC
