ARTICLE IN PRESS Clinical Nutrition (2004) 23, 341–346
www.elsevier.com/locate/clnu
ORIGINAL ARTICLE
Outcome of percutaneous endoscopic gastrostomy (PEG): comparison of two policies in a 4-year experience Galia Abuksis, Meli Mor, Shlomit Plaut, Gerald Fraser, Yaron Niv* Department of Gastroenterology, Rabin Medical Center, Beilinson Campus, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel Received 15 May 2003; accepted 13 August 2003
KEYWORDS PEG; Nutrition; Endoscopy; Gastrostomy
Summary Background: Percutaneous endoscopic gastrostomy (PEG) is the technique of choice for long-term enteral nutrition. Though safe and technically simple, PEG has been associated with significant morbidity and mortality. Aim: We compared the outcome of strategies applied in two different periods; the original approach of PEG insertion during hospitalization (upon request), and PEG insertion 30 days after hospital discharge. Methods: A cohort of 127 patients scheduled for PEG insertion from 1.1.1997 to 31.12.2000, was evaluated. In 61 consecutive patients admitted from 1.1.1997 to 31.12.1998 the PEG insertion was planned during hospitalization, as close to the time of the physician’s request (period 1). Sixty-six consecutive patients admitted from 1.1.1999 to 31.12.2000 were scheduled for the PEG insertion 30 days after discharge (period 2). The 30-day mortality rate was calculated from the time of the request. Univariate and multivariate analyses were used to find predictive factors for 30-day mortality. Results: There were 61 patients with a mean age of 78713 in period 1, and 66 patients with a mean age of 77.8715.5 in period 2. There was no significant difference between patients of the two periods in regard to age, sex, underlying disease, nutritional and mental status. Patients received PEG 30 days after hospital discharge had a 40% lower 30-day mortality rate than patients who received PEG during hospitalization from the time of request for PEG (P ¼ 0:01) and a 87.5% lower rate when calculated from the time of insertion (Po0:0001). In-hospital PEG insertion, bed-ridden and disorientation were found to be independent factors predictive of 30-day mortality after PEG insertion (P ¼ 0:016; P ¼ 0:001; and P ¼ 0:0005; respectively). Conclusion: PEG insertion during hospitalization increases mortality and should be avoided. A grace period of 30 days with nasogastric tube feeding before PEG insertion may prevent mortality and achieve a long-term enteral nutrition. & 2003 Elsevier Ltd. All rights reserved.
*Corresponding author. Tel.: þ 972-3-937-7237; fax: þ 972-3921-0313. E-mail address:
[email protected] (Y. Niv). S0261-5614/$ - see front matter & 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.clnu.2003.08.001
ARTICLE IN PRESS 342
Introduction Nutritional support is frequently needed in geriatric patients. Percutaneous endoscopic gastrostomy (PEG) is the technique of choice for long-term enteral nutrition.1–3 Though described as safe and technically simple,4–6 PEG insertion has been associated with significant morbidity and mortality.7–12 Factors such as old age (475 years), previous aspiration, urinary tract infection, dementia, hospitalization and lower body mass index increase the risk.7–14 Furthermore, it may fail to induce significant improvement in nutritional status, as noted in a 12-months study of patients with hypoalbuminemia.15 Early PEG insertion was found rewarding for patients with acute stroke or with head and neck cancer.16 Thirty-day mortality rates after PEG insertion range from 4% to 54%.3,7,11,13 There is considerable controversy in the medical literature about the indications for PEG insertion or nutritional support in different stages of disease, and the decision making process.16–19 Ethical and emotional aspects also contribute to this uncertainty.20–23 Physicians who prescribe PEG tend to focus on the patient’s inability to eat independently, and may sometimes ignore a critical underlying disease or terminal situation. It is not rare that the patient dies before the procedure takes place, or even on the same day of the request for PEG. To clarify this issue, several groups have adopted tentative policies based on quality of life or estimated life expectation.7,11,19 Prompted by our earlier study showing that PEG fed hospitalized patients had a very high 30-day mortality rate,7 we introduced a new policy of scheduling PEG for 30 days after hospital discharge date in patients for whom this procedure was prescribed. This strategy was also in line with the 30-day limit on nasogastric tube feeding effective in all Israeli nursing homes. In the present study, we looked at the outcome of the two strategies of different timing of PEG insertion.
G. Abuksis et al.
lated from the time of the request for PEG (intention to treat analysis), and from the time of actual tube insertion (per protocol). All candidates for PEG were evaluated by a gastroenterologist for general condition; nutritional status, presence of cardiopulmonary disease, and cognitive ability (correct estimation of self, place and time). Informed consent to undergo PEG was obtained in all cases from the patient or court-appointed guardian. The PEG was inserted as previously described.7 We used the same PEG kit (MIC kitFBallard Medical), and the same technique in the two periods. The department dietitian took an active part in the nutritional program in both periods. The following data were collected: demographic data, admission diagnosis, date of PEG request, indication for PEG, medical history, results of laboratory tests, date and cause of death, and date of admission. The 30-day mortality rate was calculated from 2 points: from the time of the request for PEG, and from the time of PEG insertion.
Statistical analysis Statistical analysis of data was performed using SAS software package (2001). Pearson correlation coefficient (r) and the level of significance for it (P) were calculated between variables. P values less than or equal to 0.05 were considered statistically significant. w2 test or Fisher’s exact test was used to analyze statistically significant relationships in the distribution of categorical variables. Kaplan–Meier survival curves were calculated for the cohort starting at 2 time point: since PEG request (intention to treat evaluation) and since PEG insertion (per protocol). Parameters were compared statistically between periods deferred by PEG insertion strategy and positive/negative 30-day mortality results with comparison of proportions for univariate and multivariate analysis.
Methods A cohort of 127 patients scheduled for PEG insertion in 4 years, 1.1.1997–31.12.2000, was evaluated. In 61 consecutive patients admitted from 1.1.1997 to 31.12.1998 the PEG insertion was planned during hospitalization, as close to the time of the physician’s request. Sixty-six consecutive patients admitted from 1.1.1999 to 31.12.2000 were scheduled for the PEG insertion 30 days after discharge. The 30-day mortality rate was calcu-
Results Demographic data and indications for PEG The patient characteristics are shown in Table 1. There were no significant differences between the groups in regard to age, sex, underlying disease and nutritional or mental status. Most of the patients were admitted to departments of internal medicine
ARTICLE IN PRESS New approach to PEG
Table 1
343
Patient characteristics, n(%). PEG during hospitalization (n ¼ 61)
PEG 30 days after discharge (n ¼ 66)
Gender Men Women
34 (55.8) 27 (44.2)
32 (48.5) 34 (51.5)
Age Mean7SD Range
78.0713.0 39–100
77.8715.5 40–96
Admitting department Internal medicine Neurology Oncology Otorhinolarynology
42 (68.8) 6 (9.8) 4 (6.5) 9 (14.7)
48 (72.2) 9 (13.6) 7 (10.6) 2 (3.0)
Albumin gr/dl (Mean7SD)
2.970.70
2.870.73
Table 2
Indication for PEG, n(%).
Indication
PEG during hospitalization (n ¼ 61)
PEG 30 days after discharge (n ¼ 66)
Dysphagia Previous aspiration Refuse eating due to dementia Malignancy Trauma of the head or neck
29 (47.5) 21 (34.42) 21 (34.42) 9 (14.75) 3 (4.91)
28 (42.4) 24 (36.3) 24 (36.4) 12 (18.2) 5 ( 7.6)
and had chronic diseases. The main diagnosis for the present admission was malnutrition. Prior to the request for PEG, all patients were fed orally or via a nasogastric tube. The most common indications for PEG were dysphagia or abnormal deglutition due to a cerebrovascular accident (CVA), previous aspiration, refusal to maintain adequate oral intake due to dementia, and temporary feeding in patients with cancer or trauma of the head and neck (Table 2). Eighty-five percent of the patients in the first period and 80% of the second had cognitive impairment and poor functional status; the small number of patients who were able to provide their own informed consent for PEG also reflected this. The time from admission to the date of PEG request ranged from 1 day to 1 month. This interval was significantly greater in the second period of the cohort than in the first (7.0076.82 days and 16.64718.91 days, respectively, Po0:0001). There was a significant negative correlation between this interval and 30-day mortality, the shorter the time interval for the PEG request, the higher the mortality (P ¼ 0:0027).
Hypoalbuminemia and anemia were found in most of the patients, but there was no correlation between the results of any laboratory test and mortality rate. No complications of interim feeding or inadequate nutritional supply reported in the patients of second period that were fed by a nasogastric tube.
Mortality The most common causes of death were also similar in both periods: sepsis (63%), cardiac arrest (18%) and respiratory failure (10%). The 30-day mortality from the time of the request for PEG was significantly higher in the first period than in the second (P ¼ 0:01); the survival curve is shown in Fig. 1. More than two times as many patients in the first period than of the second underwent PEG (Po0:0001); 27.8% of the former and 0.06% of the latter underwent PEG during hospitalization (Po0:0001). In the second period there were four patients who had the procedure during hospitalization because of trauma or localized cancer, in line
ARTICLE IN PRESS 344
G. Abuksis et al.
1.2
Proportion surviving
1
First period Second period
0.8
0.6
0.4
0.2
0 0
30
90
150
210
270
330
390
450
510
Days Figure 1 Survival curve since PEG request date (intention to treat analysis).
with Norton recommendation.16 Fifty-one patients (40%) underwent PEG in hospital, 68 (54%) patients were bed-ridden and 53 (42%) of the cohort were disoriented. These were found, on a multivariate regression analysis, to be independent factors predictive of 30-day mortality after PEG insertion (P ¼ 0:016; P ¼ 0:001; and P ¼ 0:0005; respectively). Twenty-eight (42.5%) patients of in the second period reached the scheduled date for PEG and 20 underwent the procedure. In the other 8 PEG was canceled because of family refusal (three patients), resumption of spontaneous eating (three patients), upper gastrointestinal bleeding (one patient), and fever (one patient). Of the 20 patients who successfully underwent PEG insertion, three died within 30 day of PEG insertion and the others survived for a mean follow-up period of 292.97175.5 days. In the first period, 43 patients (70.5%) underwent PEG; 22 died within 30 days after PEG insertion and 21 survived for a mean follow-up period of 162.867202.09 days. The 30day survival curve from the time of PEG insertion is presented in Fig. 2 (Table 3).
Discussion Primum Non Nocere is a corner stone of medicine. Although PEG may provide benefit in some patients,1–3 it is an invasive procedure associated with risks and complications,5–8 and it should be considered an option for the administration of fluids and drugs only when other routs fail or are
unfeasible.22 Most of the patients in our cohort were admitted with a main diagnosis of malnutrition and were in the terminal phase of a rapidly progressive and incurable disease. None lived long enough to benefit from the procedure, as shown by the high 30-day mortality from admission. In our previous, retrospective study, ambulatory patients were found to survive longer after PEG insertion than hospitalized patients.7 We assumed that the active disease that required hospitalization lowered survival in the hospitalized group. Therefore, we hypothesized that the mortality rate would decrease if PEG was postponed until after recovery from the acute disease plus an additional 30-day survival period with nasogastric tube feeding. The 30-day mortality rate after PEG insertion was significantly lower in period 2 than in period 1. 0ne explanation for this finding is that many patients die (with or without PEG) due to the evolution of the underlying disease. Another explanation may be improvement of general health in the grace period of 30 days before interventional procedure such as PEG. Since the nutritional status (the same average serum albumin levels) was identical in both periods, it may not have an impact on mortality after PEG insertion. A prospective, randomized, controlled study is not feasible in such a high percentage of demented patients, thus, we compared the outcome of the new policy with the original one from an earlier period. Our results showed a 40% lower 30-day mortality rate from the time of the request for PEG, and a 87.5% lower rate from the time of PEG insertion in the patients in whom the procedure was postponed
ARTICLE IN PRESS New approach to PEG
345
1.2 First period Second period
Proportion surviving
1
0.8
0.6
0.4
0.2
0
0
30
90
150
270
210
330
390
450
510
Days Figure 2 Survival curve since PEG insertion (per protocol analysis).
Table 3
Thirty-day mortality measured from three time points, n(%).
30-day mortality
PEG during hospitalization (n ¼ 61)
PEG 30 days after discharge (n ¼ 66)
P
Since admission Since request for PEG Since PEG insertiona
30 (49.18) 37 (60.65) 22 (36.06)
23 (34.84) 24 (36.36) 3 ( 4.50)
NS 0.010 o0.0001
a
For 51 patients who underwent PEG insertion.
compared to those in whom it was not (P ¼ 0:01 and Po0:0001), respectively. Thus, PEG insertion during hospitalization apparently increases mortality, it is noteworthy important to note that there was no statistically significant difference in 30-day mortality rates among patients in these two periods that were measured from the time of admission. Moreover, eight and ten patients in the first and second periods, respectively, died on the day of admission. This high mortality rate is characteristic of very high-risk, malnourished patients, and supports our new, more conservative approach. This approach is also supported by the significant negative correlation between the time interval from admission to the date of PEG request and 30-day mortality. Patients from the earlier group had PEGs requested an average of 9.6 days sooner after admission to the hospital than those from the later group. This suggests a change over time in selection criteria of the physicians requesting PEG. These patients should be fed, when needed, via nasogastric tube, until they show a significant improvement enabling hospital discharge, and an established recovery. We found in univariate and multivariate analysis that
bed-ridden, confusion and in-hospital insertion of PEG were predictive factors for mortality. Sanders et al described a high mortality rate after PEG in hospitalized demented patients.13 This finding has not been confirmed in our previous paper,7 since in ambulatory demented patients the mortality rate was lower. Thus, we believe that dementia is not an independent predictive factor of mortality, but the combination of acute illness and dementia. In support of our approach is the finding of Morrison and Siu23 that described a higher mortality rates among demented patients with pneumonia or hip fracture, than among such patients without dementia. The decision to insert PEG in senile, disabled patients is often based on emotions. In most cases it is not the patient himself who consents but a surrogate family member or a legal guardian. The consent should be based on evidence, thus the risk of PEG insertion should be stressed. For these reasons, and for moral and ethical considerations,19–24 we propose that a 30-day grace period after discharge, before PEG insertion, may prevent mortality and achieve the purpose of PEG, which is to supply long term-enteral nutrition.
ARTICLE IN PRESS 346
References 1. So JB, Ackroyd FW. Experience of percutaneous endoscopic gastrostomy at Massachusetts General Hospital, indications and complications. Singapore Med J 1998;39:560–3. 2. Nicholson FB, Korman MG, Richardson MA. Percutaneous Endoscopic gastrostomy: a review of indications, complications, and outcome. J Gastroenterol Hepatol 2000;15: 21–5. 3. Moller P, Lindberg CG, Zilling T. Gastrostomy by various techniques: evaluation of indications, outcome, and complications. Scand J Gastroenterol 1999;34:1050–4. 4. Tan W, Rajnakova A, Kum CK, Alponat A, Goh PM. Evaluation of percutaneous endoscopic gastrostomy in a university hospital. Hepatogastroenterology 1998;45:2060–3. 5. Safadi BY, Marks JM, Ponsky J. Percutaneous endoscopic gastrostomy. Gastrointest Endosc Clin North Am 1998;8:551–68. 6. Yim HB, Kaushik SP, Lau TC, Tan CC. An audit of percutaneous endoscopic Gastrostomy in general hospital in Singapore. Eur J Gastroenterol Hepatol 2000;12: 183–6. 7. Abuksis G, Mor M, Segal N, Fraser G, Niv Y. Percutaneous endoscopic gastrostomy: high mortality rate in hospitalized patients. Am J Gastroenterol 2000;95:128–32. 8. Fisman DN, Levy AR, Gifford DR, Tamblyn R. Survival after percutaneous endoscopic gastrostomy among older residents of Quebec. J Am Geriatr Soc 1999;47:349–53. 9. Finocchiaro C, Galletti R, Rovera G, et al. Percutaneous endoscopic gastrostomy: a long-term follow-up. Nutrition 1997;13:520–3. 10. Amann W, Mischinger HJ, Berger A, et al. Percutaneous endoscopic gastrostomy (PEG) 8 years of clinical experience in 232 patients. Surg Endosc 1997;11:741–4. 11. Light VL, Slezak FA, Porter JA. Predictive factors for early mortality after percutaneous endoscopic gastrostomy. Gastrointest Endosc 1995;42:330–5. 12. Friedenberg F, Jensen G, Gujral N, Braitman LE, Levine GM. Serum albumin is predictive of 30-day survival after percutaneous endoscopic gastrostomy. Parenter Enteral Nutr 1997;21:72–4.
G. Abuksis et al.
13. Sanders DS, Carter MJ, D’Silva J, et al. Percutaneous endoscopic gastrostomy: a prospective audit of the impact of guidelines in two district general hospitals in the united kingdom. Am J Gastroenterol 2002;97:2239–45. 14. Nair S, Hertan H, Pitchumoni CS. Hypoalbominemia is a poor predictor of survival after percutaneous endoscopic gastrostomy in elderly patients with dementia. Am J Gastroenterol 2000;95:133–5. 15. Kaw M, Seaks G. Long-term follow-up of consequences of percutaneous endoscopic gastrostomy (PEG) tubes in nursing home patients. Dig Dis Sci 1994;39:738–40. 16. Norton B, Homer MW, Donnelly MT, Long RG, Holmes GKT. A randomized prospective comparison of percutaneous endoscopic gastrostomy and nasogastric tube feeding after acute dysphagic stroke. Br Med J 1996;312:13–6. 17. Rickman J. Percutaneous endoscopic gastrostomy: psychological effects. Br J Nurs 1998;7:723–9. 18. Callahan CM, Haag KM, Buchanan N, Nisi R. Decision-making for percutaneous endoscopic gastrostomy among older adults in a community setting. J Am Geriatr Soc 1999;47:1105–9. 19. Van Rosendaal GM, Verhoef MJ, Mace SR, Kinsella TD. Decision-making and outcomes for percutaneous endoscopic gastrostomy: a pilot study. J Clin Gastroenterol 1997;24: 71–3. 20. Rabeneck L, Wray NP, Petersen NJ. Long term outcomes of patients receiving percutaneous endoscopic gastrostomy tubes. J Gen Int Med 1996;11:287–93. 21. Rosin AJ, Sonnenblick M. Autonomy and paternalism in geriatric medicine. The Jewish ethical approach to issues of feeding terminally ill patients and to cardiopulmonary resuscitation. J Med Ethics 1998;24:44–8. 22. Rabeneck L, McCullough LB, Wray NP. Ethically justified, clinically comprehensive guidelines for percutaneous endoscopic gastrostomy tube placement. Lancet 1997;349:496–8. 23. Morrison RS, Siu AL. Survival in end-stage dementia following acute illness. JAMA 2000;284:47–52. 24. Guido MA, Van Rosendaal GM, Verhoef MJ, Kinsella TD. How are decisions made about the use of percutaneous endoscopic gastrostomy for long term nutritional support? Am J Gastroenterol 1999;94:3225–8.
