World J Surg (2015) 39:2557–2563 DOI 10.1007/s00268-015-3115-4

ORIGINAL SCIENTIFIC REPORT

Determinants of Surgical Site Infections Following Pancreatoduodenectomy Savio George Barreto1 • Manish Kumar Singh2 • Sunil Sharma3 • Adarsh Chaudhary1

Published online: 10 June 2015  Socie´te´ Internationale de Chirurgie 2015

Abstract Background Surgical site infections (SSI) following pancreatoduodenectomy (PD) contribute to adverse perioperative and long-term outcomes. Hence, the need to determine the modifiable factors related to their causation. Aim To identify demographic, nutritional, surgical and histopathological factors significantly associated with incisional SSIs. Methods A retrospective analysis of a prospectively maintained database of consecutive patients who underwent PD for pancreatic and periampullary lesions at a tertiary care referral centre, between April 2010 and June 2014 was carried out. Patients were divided into two groups based on the SSIs (Group 1—With SSI; Group 2—No SSI). All patients were administered three, once daily doses of Ertapenem (1 g) as follows: within 1 h prior to induction, and on day 1 and day 2 following surgery. No further antibiotics were given prior to discharge unless clinically indicated. Results 35 out of 277 patients (12.6 %) developed SSIs. No demographic (age, sex, BMI), nutritional (serum albumin), surgical (pancreatic duct size and texture, surgical duration and intraoperative blood transfusions) and histopathological factors (malignancy vs. benign) were noted between the two groups. However, SSIs were significantly higher in patients with endocrine co-morbidities (other than diabetes mellitus), in those patients who had undergone prior ERCP and stenting, as well as an end-to-side pancreaticojejunostomy. Patients with diabetes mellitus had a significantly lower incidence of wound infections (P = .014). Conclusion Preoperative ERCP and stenting, end-to-side PJ and the presence of non-diabetic endocrine co-morbidity may result in a significantly higher risk of SSIs. Further studies targeting these patient subpopulations are warranted to enable a better understanding of how these factors contribute to the incidence of SSIs following PD.

Introduction & Adarsh Chaudhary [email protected] Savio George Barreto [email protected] 1

Department of Gastrointestinal Surgery, Gastrointestinal Oncology, and Bariatric Surgery, Medanta Institute of Digestive and Hepatobiliary Sciences, Medanta, The Medicity, Sector 38, Gurgaon, Haryana, India

2

Department of Clinical Research, Medanta, The Medicity, Gurgaon, India

3

Department of Microbiology, Medanta, The Medicity & Current Dean, Indian Spinal Injuries Centre, Gurgaon, India

Pancreatoduodenectomy (PD), or the Whipple’s operation, is associated with major morbidity and even the risk of mortality [1]. One of the most commonly reported complications of PD is surgical site infections (SSIs). SSIs following PD have been reported to vary from 7 to 32 % [2, 3] and contribute to increased length of hospital stay [4] and even hospital re-admissions [5] leading to escalating costs. However, SSIs may assume a more sinister perspective in their ability to result in mortality [6, 7]. Besides, the presence of surgical complications such as SSIs precludes the initiation of chemotherapy in patients

123

2558

undergoing surgery for cancer [8]. This can potentially influence the patients overall survival [9]. Thus, a concerted effort at reducing SSIs following PD is needed. If the incidence of SSIs is to be reduced, it is important to determine the specific factors responsible for their causation following PD rather than extrapolating data from other organ surgeries. Donald et al. [3] demonstrated a significant reduction in SSIs following PD by simply changing the perioperative antibiotic prophylaxis from the existing colonic/abdominal surgery protocols to a regimen tailored for PD. From literature, some of the factors attributed to the development of SSIs following pancreatic surgery include malnutrition and hypoalbuminemia [10–12], preoperative biliary drainage [13, 14] and surgery-related factors such as the development of length of surgery [15], main pancreatic duct diameter [15], pancreatic fistula, post-operative bleeding, long-term ascites and re-operation [4, 15]. The aim of the present study was to identify demographic, nutritional, surgical and histopathological factors significantly associated with SSIs following PD. The determination of such factors may enable us to develop strategies targeting such ‘at risk’ patients thereby potentially reducing the incidence of SSIs following PD.

Patients and methods Patient population The data of 277 consecutive patients who underwent PD for pancreatic and periampullary lesions (benign and malignant) at Medanta, the Medicity, a tertiary care referral centre, between April 2010 and June 2014 were obtained from a prospectively maintained database. Study groups The patient cohort was divided into two groups, based on the presence of incisional SSIs, viz: • •

Group 1—With SSI, and Group 2—No SSI.

Choice of antibiotics All the 277 patients were administered three, once daily doses of Ertapenem (1 gram) as follows: within 1 h prior to induction, and on day 1 and day 2 following surgery. Antibiotics were then ceased following post-operative day 2 until discharge. Therapeutic use of antimicrobials, including a change in antibiotics, was considered in patients who thereafter developed infectious complications based on the site of infection and appropriate culture and sensitivity report.

123

World J Surg (2015) 39:2557–2563

Ethics The study has been approved by the Institutional Review Board and the Ethics Committee of the Medanta, The Medicity (Approval No: MICR-346/2013). The post-PD clinical pathway for post-operative care followed in our Unit has been previously published [16]. All patients underwent a classical PD, after an overnight fast, via a right subcostal incision with the following reconstruction: Pancreaticojejunostomy (PJ)—end to side, duct to mucosa in two layers, hepaticojejunostomy—end to side, gastrojejunostomy—side to side, retrocolic, retrogastric. In patients in whom the pancreatic duct was not visualised, an end-to-side pancreaticojejunostomy was performed. A standard lymphadenectomy was performed in all patients. The decision on whether or not to place a single drain (32 Fr Portex drain in the Morisson’s pouch) intraoperatively was made on a case-by-case basis. An Ioban  wound barrier system was routinely applied. Definition of complications For the study, infectious complications were defined as SSIs and/or intraabdominal collections or respiratory infective complications associated with signs of sepsis (transient or prolonged) including fever, elevated leucocytosis and the need for re-initiation of antibiotics. SSIs were defined as per the Centres for Disease Control definition (1992). Post-operative pancreatic anastomotic leak/fistula (POPF), delayed gastric emptying (DGE), bile leak and postpancreatectomy haemorrhage (PPH) were defined as per the International Study Group of Pancreatic Surgery (ISGPS) criteria [20–22]. Mortality was defined as any death that has occurred not only in the 30 days following surgery but even thereafter so long as there exists sufficient evidence to link the cause of death to the surgery or its complications. Any readmissions within 30 days of discharge, whether in our hospital or another hospital, were recorded. Statistical analysis The analysis includes profiling of patients on different demographic and clinical as well as treatment outcome parameters. The term, ‘non-diabetic endocrine co-morbidity’ includes thyroid, adrenal and pituitary diseases in patients without diabetes mellitus. Descriptive statistics of quantitative variables are expressed as means and standard deviation. Ordinal/categorical data are expressed in terms of proportions and percentages. The assessment of normality of data was tested using the Kolmogorov–Smirnov test. For the comparison of means between groups, Independent Student t test was performed. Standard normal deviate test (Z test) and v2 test have been used for the

World J Surg (2015) 39:2557–2563

2559

Table 1 Comparison of demographic and surgical variables between the two patient groups Parameter

Group 1 (n = 35)

Group 2 (n = 242)

P value

Mean age in years (±SD)

55.8 (±11.8)

56.4 (±13)

0.795

Sex (Male:female)

23:12

170:72

0.585

Mean BMI in kg/m2 (±SD)

24.5 (±3.3)

24.5 (±4.6)

0.980

Mean serum albumin in g % (±SD)

3.6 (±0.7)

3.4 (±0.7)

0.151

19 (54.3) 16 (45.7)

135 (56) 107 (44)

22 (63)

131 (54)

Patient Demography

Surgical factors Pancreatic texture (%) Normal/soft Firm

0.900

Pancreatic duct diameter (mm) \3

0.518

3–5

11 (31)

85 (35)

[5

2 (6)

26 (11)

0

7 (20)

54 (22)

0.170

1

28 (80)

188 (78)

0.285

Surgical (interoperative) drains (%)

0.717

Mean duration surgery in minutes (±SD)

308 (±58)

294 (±54)

Mean interoperative blood loss in ml (±SD)

245 (±128)

219 (±139)

SD standard deviation, BMI body mass index

qualitative variables. Multivariate logistic regression was performed by comparing variables carrying univariate associations with the outcome and a P value (\0.05). P value \0.05 is considered statistically significant. All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) INC., Version 18.0 for Windows.

Results Patient demography The cohort of 277 patients consisted of 193 male and 84 female patients with a median age of 58 years (range 15–80), median BMI of 24.3 kg/m2 (range 12.5–42.5) and median serum albumin level of 3.5 g % (range 1.3–5). The comparative demographic data of the 277 patients in the two groups are presented in Table 1. There were a total of 35 patients in Group 1 and 242 patients in Group 2, respectively. No significant difference in the parameters was noted between the two groups. Associated co-morbidities and prior interventions (ERCP and stenting) Table 2 provides the comparison between the two groups in terms of pre-existing co-morbidities and American Society of Anaesthesiologists (ASA) grade as well as prior

ERCP and stenting. Patients with non-diabetic endocrine co-morbidities and those who underwent ERCP with stenting were found to have a significantly higher incidence of wound infections. Surgery-related factors All patients underwent a classical Whipple’s procedure with a median duration of 290 min (range 205–480). Surgical factors including pancreatic texture, duct size, use of drains, blood loss, as well as duration of surgery are compared between the two groups in Table 1. There was no significant difference between the two groups with regard to these parameters. The need for intraoperative blood transfusions (with its attendant risk of SSI) was also compared and found to be not significant (P = .732). However, patients who underwent an end-to-side PJ were found to be significantly higher in group 1 (14 vs. 5.8 %; P = .028) (Table 2). Surgical outcomes The overall morbidity and mortality rates for the entire cohort were 30.7 and 4.3 %, respectively. Table 3 provides a comparison between the two groups with regard to surgical and medical co-morbidities and POPF rates. As per the ISGPF definition, the grades of POPF were as follows: Grade A = 2 patients, Grade B = 4 patients and Grade C = 1 patient. No significant difference was noted in these variables between the two groups.

123

2560

World J Surg (2015) 39:2557–2563

Table 2 Comparison of preoperative interventions (ERCP and stenting), pre-existing co-morbidities and intraoperative factors between the two groups Parameters

Study Group

P value

Wound infection 35 (%)

No wound infection 242 (%)

Yes

23 (65.7)

90 (37.3)

No

12 (34.3)

151 (62.7)

Yes

23 (65.7)

86 (35.7)

No

12 (34.3)

155 (64.3)

Preoperative factors ERCP 0.001*

Stent 0.001*

Co-morbidities American Society of Anaesthesiologists grade I

9 (25.7)

71 (29.3)

II

22 (62.9)

148 (61.2)

0.874

III

4 (11.4)

23 (9.5)

5 (14.3) 30 (85.7)

85 (35.1) 157 (64.9)

0.014*

Yes

14 (40.0)

86 (35.5)

0.607

No

21 (60.0)

156 (64.5)

Diabetes mellitus Yes No Hypertension

Cardiovascular system Yes

6 (17.1)

35 (14.5)

No

29 (82.9)

207 (85.5)

0.676

Renal Yes

0 (0.0)

2 (0.8)

No

35 (100.)

240 (99.2)

0.589

Central Nervous System Yes

0 (0.0)

6 (2.5)

No

35 (100.0)

236 (97.5)

0.346

Non-diabetic endocrine diseases (including thyroid and pituitary) Yes No

5 (14.3) 30 (85.7)

10 (4.1) 232 (95.9)

0.013*

0.589

Cirrhosis Yes

0 (0.0)

2 (0.8)

No

35 (100.0)

240 (99.2)

Yes

3 (8.6)

21 (8.7)

No

32 (91.4)

221 (91.3)

Portal Vein resection (incl. lateral wall wedge)

1 (2.9)

5 (2.1)

SMV with graft

0 (0.0)

1 (0.4)

No

34 (97.1)

235 (97.5)

Respiratory 0.983

Intraoperative factors Vascular resection 0.891

Pancreatic anastomosis (PJ) Duct to mucosa

30 (85.7)

227 (94.2)

End to side

4 (11.4)

13 (5.4)

Side to side

1 (2.9)

1 (0.4)

123

0.028*

World J Surg (2015) 39:2557–2563

2561

Table 2 continued Parameters

Study Group

P value

Wound infection 35 (%)

No wound infection 242 (%)

0

31 (93.9)

207 (90.4)

1 2

1 (3.0) 1 (3.0)

15 (6.6) 7 (3.1)

Cancer

30 (88.2)

213 (88.0)

Benign

4 (11.8)

29 (12.0)

Intraoperative blood transfusions 0.732

Post-operative factors Underlying pathology 0.971

* P value \0.05-statistically significant ERCP Endoscopic retrograde cholangiopancreatography, SMV superior mesenteric vein, PJ Pancreaticojejunostomy

Table 3 Post-operative outcomes in the two groups Group 1 (n = 35)

Group 2 (n = 242)

P value

Surgical Complications

6 (17 %)

79 (33 %)

.063

Medical complications

1 (3)

18 (7)

.316

POPF

0 (0)

7 (2.9)

.308

Pathological factors The presence of cancer, as the underlying pathology, did not significantly affect the development of SSIs (P = .971) (Table 2). Factors influencing SSIs following PD The four factors determined on univariate analysis to be significantly associated with the causation of SSIs were tested in a multivariate model. The results are summarised in Table 4. The presence of non-diabetic endocrine co-morbidity (OR 5.494), preoperative ERCP and stenting (OR 4.561) were independent risk factors for SSIs. Similarly, patients with diabetes mellitus (OR 0.220) or the performance of a duct-to-mucosa PJ (OR 0.183) were independently found to be less likely to be associated with the development of SSIs.

Discussion These data from one of the largest cohorts of patients undergoing PD specifically addressing SSIs indicate that preoperative ERCP and stenting, end-to-side PJ and the presence of non-diabetic endocrine co-morbidity may result in a significantly higher risk of SSIs. Patients with diabetes mellitus were found to have significantly lower incidence of SSIs.

SSIs account for one of the more common complications following PD. They contribute to increased length of hospital stay [4], hospital re-admissions [5] and thus lead to escalating medical costs. SSIs may also result in mortality [6, 7] and impact overall survival [9] by causing delays in initiating adjuvant therapy in patients undergoing surgery for cancer. A mere extrapolation of regimens from other organ systems may not be enough as has been demonstrated by Donald et al. [3]. Thus, the present study attempted to determine the risk factors associated with SSIs in a large cohort of patients undergoing PD by comparing the demographic and nutritional (age, sex, BMI and serum albumin levels), surgical (pancreatic duct size and texture, use of drains, intraoperative blood transfusions, duration of surgery) and histopathological (benign and malignant disease) factors, which are known to affect outcomes following PD. In this study, patients who underwent preoperative ERCP and stenting were found to be more likely to develop SSIs. This finding has been noted in other studies as well [17–19]. Biliary colonisation is a direct off shoot of this practice (as has been noted in this study and by others [19]) with a large number of patients going on to develop complications such as cholangitis and acute pancreatitis. These have been shown to influence the incidence of surgical complications [17, 20, 21]. A randomised controlled trial is currently underway exploring the role of wound protectors in preventing SSIs in patients undergoing PD after preoperative biliary drainage [22].

123

2562

World J Surg (2015) 39:2557–2563

Table 4 Multivariate analysis of risk factors influencing SSIs P value

Odds ratio

95 % confidence interval

Diabetes Mellitus

0.005*

0.220

0.076–0.637

Non-diabetic endocrine co-morbidity

0.009*

5.494

1.540–19.596

ERCP and stenting

0.000*

4.561

2.000–10.402

Duct-to-Mucosa PJ

0.006*

0.183

0.055–0.614

* P value \0.05-statistically significant PJ Pancreaticojejunostomy

Although the data failed to show a significant difference between the two groups in pancreatic texture (P = .90) and main pancreatic duct size (P = .51), SSIs were significantly more common in patients who underwent an end-toside PJ. We tend to favour the performance of an end-toside PJ over a duct-to-mucosa PJ in patients with a small main pancreatic duct diameter (B2 mm). While evidence in surgical literature supports a higher incidence of POPF in patients with small pancreatic duct diameters [23], we were unable to document an association between the development of POPF and SSIs (P = .308). In our cohort of patients, SSIs were more frequently encountered in those with endocrine co-morbidities (other than diabetes mellitus). This finding has not been previously reported following PD. However, the endocrine stress response to surgery has been well documented, mediated by an increase in pituitary hormone secretion and the activation of the sympathetic nervous system [24]. Such a stress response was noted at 1 and 24 h following ERCP in a study by Adas et al. [25] where significant elevations in adrenocorticotropins, cortisol, growth hormone and aldosterone were noted. A stress response may result in metabolic alterations, one of which is hyperglycaemia, contributing to an increased risk of SSIs. Our data failed to demonstrate a significant association between demographic and nutritional (age, sex, BMI and serum albumin levels), surgical (pancreatic duct size and texture, use of drains, intraoperative blood loss and transfusions, duration of surgery) and histopathological (benign and malignant disease) factors and the development of SSIs as has been shown by others [10–15, 26, 27]. Similarly, there was no significant difference between the two groups in terms of the surgical or medical morbidity. Another, interesting finding in our study has been the significantly lower incidence of SSIs in diabetics as compared to non-diabetics. This is in contrast to other studies which demonstrate a higher rate of SSIs in diabetics possibly related to problems with their glycaemic control in the perioperative period [28]. Since the study was not specifically designed to address this cohort of patients, we are unable to readily provide an explanation for this observation. However, we could opine that this could

123

reflect a better glycaemic control in the perioperative phase in our patients. The biggest strength of the present study was the large size of the cohort. Previous studies looking at SSIs following PD, other than the study by Suguira et al. [15], have involved smaller cohorts of patients [3, 4, 10, 12, 29]. Small sample sizes are fraught with the risk of sampling biases and a tendency to overestimate risk factors [30]. The segregation of diabetes from other endocrine co-morbidities, as opposed to clubbing them into one group, provides a novel risk factor for SSIs that can be investigated further. On the other hand, the weakness of the study is the retrospective nature of analysis which precluded a determination of the mechanisms by which the risk factors actually influenced the development of incisional SSIs. However, the study does provide certain inferences that can be pursued in order to understand the underlying pathophysiology.

Conclusions Preoperative ERCP and stenting, end-to-side PJ and the presence of non-diabetic endocrine co-morbidity may result in a significantly higher risk of SSIs. Further studies targeting these patient subpopulations are warranted to enable a better understanding of how these factors contribute to the incidence of SSIs following PD. Acknowledgments This study was awarded an Investigator Initiated Study grant by Merck Sharp & Dohme (MSD Regional Office, India) - MISP Database number 51222. Conflict of interest

None.

Commercial/financial disclosures

None.

References 1. Shrikhande S, Barreto S (2012) Surgery for pancreatic carcinoma: state of the Art. Indian J Surg 74(79–8):6 2. Ueno T, Yamamoto K, Kawaoka T et al (2005) Current antibiotic prophylaxis in pancreatoduodenectomy in Japan. J Hepatobiliary Pancreat Surg 12(304–30):9

World J Surg (2015) 39:2557–2563 3. Donald GW, Sunjaya D, Lu X et al (2013) Perioperative antibiotics for surgical site infection in pancreaticoduodenectomy: does the SCIP-approved regimen provide adequate coverage? Surgery 154:190–196 4. Nanashima A, Abo T, Arai J et al (2014) Clinicopathological parameters associated with surgical site infections in patients who underwent pancreatic resection. Hepatogastroenterology 61(1739–174):3 5. Kirkland KB, Briggs JP, Trivette SL et al (1999) The impact of surgical-site infections in the 1990s: attributable mortality, excess length of hospitalization, and extra costs. Infect Control Hosp Epidemiol 20:725–730 6. Mangram A, Horan T, Pearson M et al (1999) Guideline for prevention of surgical site infection (CDC) Hospital Infection Control Practices Advisory Committee. Am J Infect Control 27:97–132 7. Olson MM, Lee JT Jr (1990) Continuous, 10-year wound infection surveillance. Results, advantages, and unanswered questions. Arch Surg 125:794–803 8. Merkow RP, Bilimoria KY, Tomlinson JS et al (2013) Postoperative complications reduce adjuvant chemotherapy use in resectable pancreatic cancer. Ann Surg 260(2):372–7 9. Valle JW, Palmer D, Jackson R et al (2014) Optimal duration and timing of adjuvant chemotherapy after definitive surgery for ductal adenocarcinoma of the pancreas: ongoing lessons from the ESPAC-3 study. J Clin Oncol 32:504–512 10. Shinkawa H, Takemura S, Uenishi T et al (2013) Nutritional risk index as an independent predictive factor for the development of surgical site infection after pancreaticoduodenectomy. Surg Today 43(276–28):3 11. Hennessey DB, Burke JP, Ni-Dhonochu T et al (2010) Preoperative hypoalbuminemia is an independent risk factor for the development of surgical site infection following gastrointestinal surgery: a multi-institutional study. Ann Surg 252(325–32):9 12. La Torre M, Ziparo V, Nigri G et al (2013) Malnutrition and pancreatic surgery: prevalence and outcomes. J Surg Oncol 107(702–70):8 13. Cortes A, Sauvanet A, Bert F et al (2006) Effect of bile contamination on immediate outcomes after pancreaticoduodenectomy for tumor. J Am Coll Surg 202(93–9):9 14. Pisters PW, Hudec WA, Hess KR et al (2001) Effect of preoperative biliary decompression on pancreaticoduodenectomy-associated morbidity in 300 consecutive patients. Ann Surg 234:47–55 15. Sugiura T, Uesaka K, Ohmagari N et al (2012) Risk factor of surgical site infection after pancreaticoduodenectomy. World J Surg 36(2888–289):4. doi:10.1007/s00268-012-1742-6 16. Chaudhary A, Barreto S, Talole S et al (2014) Early discharge after Pancreatoduodenectomy—what helps and what prevents? Pancreas 44(2):273–8

2563 17. Morris-Stiff G, Tamijmarane A, Tan YM et al (2011) Pre-operative stenting is associated with a higher prevalence of postoperative complications following pancreatoduodenectomy. Int J Surg 9(145–14):9 18. Sivaraj SM, Vimalraj V, Saravanaboopathy P et al (2010) Is bactibilia a predictor of poor outcome of pancreaticoduodenectomy? Hepatobiliary Pancreat Dis Int 9:65–68 19. Sourrouille I, Gaujoux S, Lacave G et al (2013) Five days of postoperative antimicrobial therapy decreases infectious complications following pancreaticoduodenectomy in patients at risk for bile contamination. HPB (Oxford) 15:473–480 20. Ngu W, Jones M, Neal C et al (2013) Preoperative biliary drainage for distal biliary obstruction and post-operative infectious complications. ANZ J Surg 83(280–28):6 21. Yanagimoto H, Satoi S, Yamamoto T et al (2014) Clinical impact of preoperative cholangitis after biliary drainage in patients who undergo pancreaticoduodenectomy on postoperative pancreatic fistula. Am Surg 80:36–42 22. Bressan AK, Roberts DJ, Edwards JP et al (2014) Efficacy of a dual-ring wound protector for prevention of incisional surgical site infection after Whipple’s procedure (pancreaticoduodenectomy) with preoperatively-placed intrabiliary stents: protocol for a randomised controlled trial. BMJ Open 4:e005577 23. Shukla PJ, Barreto SG (2010) Fingerhut A Do transanastomotic pancreatic ductal stents after pancreatic resections improve outcomes? Pancreas 39:561–566 24. Desborough J (2000) The stress response to trauma and surgery. Br J Anaesth 85:109–117 25. Adas G, Kemik A, Adas M et al (2013) Metabolic and inflammatory responses after ERCP Int. J Biomed Sci 9(237–24):2 26. Gibbons C, Bruce J, Carpenter J et al (2011) Identification of risk factors by systematic review and development of risk-adjusted models for surgical site infection. Health Technol Assess 15:1–156 iii-iv 27. Ross A, Mohammed S, Vanburen G et al (2013) An assessment of the necessity of transfusion during pancreatoduodenectomy. Surgery 154(504–51):1 28. Ata A, Lee J, Bestle S et al (2010) Postoperative hyperglycemia and surgical site infection in general surgery patients. Arch Surg 145(858–86):4 29. Ambiru S, Kato A, Kimura F et al (2008) Poor postoperative blood glucose control increases surgical site infections after surgery for hepato-biliary-pancreatic cancer: a prospective study in a high-volume institute in Japan. J Hosp Infect 68(230–23):3 30. Morgan T, Coffey C (2003) Krumholz H Overestimation of genetic risks owing to small sample sizes in cardiovascular studies. Clin Genet 64(7–1):7

123