Elsevier

The Lancet

Volume 356, Issue 9242, 11 November 2000, Pages 1632-1637
The Lancet

Articles
Impact of laparoscopic cholecystectomy: a population-based study

https://doi.org/10.1016/S0140-6736(00)03156-1Get rights and content

Summary

Background

We assessed the effect of the introduction of laparoscopic cholecystectomy on surgical outcomes in routine practice.

Methods

Hospital discharge and death-certificate data were linked for all patients undergoing cholecystectomy (n=85 120) in Scottish public-sector hospitals (n=51) between January, 1981, and June, 1999. The primary endpoints were cholecystectomy rate, hospital stay, and postoperative mortality. Regression methods were used to examine the effect of laparoscopic experience and surgeon caseload on postoperative mortality and hospital stay.

Findings

From 1989 to 1999, the proportion of cholecystectomies done laparoscopically rose from none to 80%, and the age-standardised cholecystectomy rate increased by 20% (95% CI 15–26). Postoperative mortality did not change in the 1990s (odds ratio 0·99 [0·7–1·4], p=0·99). The mean postoperative hospital stay fell from 8·0 (SD 3·7) to 2·9 (3·2) days. There was wide variation between hospitals in the proportion of cholecystectomies done laparoscopically and in average hospital stay. For individual surgeons, increasing laparoscopic experience and annual caseload were associated with higher proportions of laparoscopic procedures and shorter hospital stays. Postoperative mortality was higher during the first ten laparoscopic cholecystectomies done by a surgeon (compared with >200 procedures, odds ratio 2·3 [1·2–4·6], p=0·015).

Interpretation

The laparoscopic method reduced hospital stay but had no overall effect on postoperative mortality. Studies to assess the appropriateness of the increased cholecystectomy rate are merited. The wide variation in the proportion done laparoscopically, together with evidence of better results for surgeons doing more procedures, suggests scope for further reductions in hospital stay and morbidity.

Introduction

In the early 1990s, laparoscopic cholecystectomy rapidly and almost completely replaced the open method. Subsequently, reduced postoperative pain and faster recovery were reported from randomised trials.1, 2 However, it is unclear to what extent hospital stay has been reduced in routine practice. Two large American audits (each of more than 150 hospitals) found that hospital stay fell by only 1 day3 or 2 days,4 to an average of 6–7 days.

Enthusiasm for the laparoscopic method was associated with an increase (ranging in published series from 10% to 69%) in the number of cholecystectomies done.5, 6, 7 Whether this increase has been sustained beyond the early 1990s is not known.

Laparoscopic cholecystectomy has a lower postoperative mortality than open surgery but this difference is largely explained by case selection, high-risk patients being more likely to undergo open cholecystectomy.8 Complications caused by technical errors (such as bile-duct injury, bile leak, and vascular injury) have increased since the introduction of the laparoscopic method,9, 10, 11, 12, 13 but general complications have been reduced.12 Only one study has shown an absolute reduction in the number of postoperative deaths.4 There are no population-based mortality data available after 1995 when morbidity from the learning curve should have diminished. Thus, the ultimate effect of the laparoscopic method on postoperative mortality remains unclear.

We used a population-based record-linkage database to find first the effect of the introduction of the laparoscopic method on cholecystectomy rate, operative mortality, and hospital stay, and, second, the effect of individual surgical experience and surgical workload on these variables.

Database

Summary information is recorded in the Scottish Morbidity Record 1 (SMR1) database for each patient discharged from National Health Service acute hospitals in Scotland. The office of the Registrar General records all deaths in Scotland. Individual records from these two sources are linked by probability matching to produce a single group of records for each patient, irrespective of the admitting hospital and home-address changes. The Scottish population is stable, with an annual migration of only 1–1·5% to and from other parts of the UK or overseas.14 The accuracy of the record-linkage database is ensured through continuous staff training and intensive quality assurance. An annual audit of 1% of hospital returns is done: the record-linkage process, procedure codes, and diagnostic codes have an accuracy of 99%, 94%, and 90%, respectively.15 The SMR1-linked database currently holds information from 1981 to June, 1999.

Patients

The database was searched for all patients undergoing cholecystectomy. Procedures done in the private sector (which we estimate account for less than 5% of cholecystectomies) were not included, since these hospitals were unwilling to provide information. We extracted data on age at time of operation, sex, type of cholecystectomy, exploration of common bile duct (ECBD), endoscopic retrograde cholangiopancreatography (ERCP), type of admission, length of hospital stay, operative and diagnostic codes, vital status up to June 30, 1999, hospital, and surgeon responsible. There was no separate code for procedures converted from laparoscopic to open, and it is likely that these were coded as open. Because the study involved the use of routinely collected data anonymised to avoid identification of patients, hospitals, or surgeons, it was not thought necessary to obtain approval of the ethics committee or informed consent.

Patients with a diagnosis of hepato-pancreatico-biliary malignant disease (n=1004) were excluded, leaving 89 186 patients. Cholecystectomy rates were calculated from population data estimated by the Registrar General for Scotland,14 and were directly age-standardised with the European standard population. For all other analyses, patients who underwent an unrelated major operation during the same admission (n=3718) were excluded, leaving 85 458 patients for analysis.

Mortality was increased up to 90 days after the procedure, and therefore both 30-day (the standard measure) and 90-day mortality were calculated. 1-year mortality (excluding 90-day mortality) was calculated as a measure of the general health of patients.

Statistical analysis

The primary endpoints were age-standardised cholecystectomy rate, postoperative 90-day mortality, and hospital stay. We also examined the effect of laparoscopic experience and surgeon caseload on postoperative mortality and hospital stay. We analysed the data using Access 2000, Excel 2000, SPSS 9·0 for Windows, and Stata 6·0 for Windows. Patients with long hospital stays (which are frequently due to unrelated medical or social problems) have an excessive effect on the mean hospital stay. To avoid this bias, mean postoperative hospital stay was calculated by limiting stay to the trim point of the complete patient group (third quartile plus twice the interquartile range10), which was 21 days. We used logistic regression to estimate odds ratios for postoperative mortality, and linear regression to estimate adjusted differences in postoperative stay. Postoperative death, sex, bile-duct exploration, and preperative acute pancreatitis were coded as binary variables. Indicator variables were used to code age in five categories, date of operation grouped into 2-year intervals, the cumulative number of laparoscopic cholecystectomies in seven categories, and the number of procedures done yearly by the individual surgeon in five categories. The youngest age group, the earliest year, and the largest number of procedures were taken as the reference categories. The date of operation was added as a continuous variable to the regression analyses (for comparison of laparoscopic and open cholecystectomies and analysis of influence of caseload and laparoscopic learning curve) to control for time trends during 1990–99. All variables were entered without selection into models that took account of the clustered nature of the data by the use of robust variance estimates, specifying the hospitals as the unit of clustering.

Procedures

The proportion of cholecystectomies done laparoscopically increased rapidly from 1990 to 1993, reaching 80% by 1999 (figure, A). This proportion may be as high as 83% if procedures coded as open, but with a postoperative stay of 1 day or 2 days, represent miscoding. The proportion of cholecystectomies done laparoscopically varied widely among hospitals (in 1999, range 0–100%, IQR 71–88%). By 1999, 90% of operations were done by surgeons who had done more than 30 laparoscopic cholecystectomies.

The age-standardised cholecystectomy rate declined from 95 to 83 procedures per 100 000 population in the decade before the introduction of laparoscopic cholecystectomy (figure, B). The fall in 1982 was due to industrial action in the National Health Service and the rise in 1983 was caused by a waiting-list initiative. Laparoscopic cholecystectomy was introduced from 1990. The total number of procedures increased from 83 to 98 per 100 000 between 1990 and 1993 and remained at about this level till 1999 (when it reached 100). The increase from 1989 to 1999 was 20% (95% CI 15–26). The increases varied greatly by age and sex (table 1). The greatest increase (a third) was in women aged 55 to 64 years, whereas the overall increase in men was modest (3%).

The percentage of patients who underwent ERCP in the year before or after their cholecystectomy increased from 5% in 1989 to 22% in 1998 (figure, C). Although the use of bile-duct exploration fell from 11% to 3% (figure, C), 47% of patients undergoing bile-duct exploration also underwent ERCP. In 1996–98, ERCP rates in Scottish hospitals varied from 0 to 35% (IQR 15–23%). Surprisingly, there was no significant inverse correlation between ERCP and bile-duct exploration rates of each hospital (Spearman rank correlation r=−0·11, p=0·50).

Characteristics of patients

The age distribution (mean 53 years [SD 16] did not change significantly during the study period (Mann-Whitney U test, 1980s vs 1990s, p=0·21), whereas the proportion of women increased slightly (1980s 74% vs 1990s 75% difference 1% [95% CI 0·5–1·7]).

Type of admission

The proportion of all operations done during emergency admissions was 20% in the 1980s, but after laparoscopic cholecystectomy was introduced it fell to 15% (1992–96), later increasing slightly to 17% (1997–99) (figure, D). There was considerable variation between hospitals in the percentage of operations during emergency admission (in 1999, range 0–50%, IQR 6–18%).

Hospital stay

Before the introduction of laparoscopic cholecystectomy there was a reduction in postoperative hospital stay from a mean of 9·5 to 8·0 days (figure, E). There was a sharp reduction in postoperative hospital stay from 7·4 to 4·0 days during 1990–93, associated with the introduction of the laparoscopic method, and a further gradual decline to 2·9 days by 1999. There was a wide variation in the average postoperative length of stay between hospitals; in 1999, it ranged from 1·3 to 6·3 days (IQR 2·5–3·5).

From 1985 to 1996, there was a linear rise from 2·2% to 5·4% in the proportion of patients readmitted as an emergency (for any reason) within 28 days of discharge, falling slightly to 4·5% in 1999.

Postoperative mortality

For the whole group patients mortality during the first 30 days after the procedure was 0·74% (95% CI 0·68–0·80), during the second 30 days 0·23% (0·19–0·26), during the third 30 days 0·18% (0·15–0·20), and thereafter was not significantly different from the overall mean of 0·13% (0·12–0·13). We therefore used 90-day mortality as the primary postoperative mortality endpoint.

By the use of logistic regression, the significant independent risk factors in the 1990s for postoperative death within 90 days were age (odds ratio relative to age <55 years: 55–64 years 3·0 [95% CI 2·0–4·5], 65–74 years 7 [5–9], 75–84 years 13 [9–19], and >84 years 31 [20–47]), male sex (odds ratio 1·5 [1·2–1·9]), emergency admission (4·9 [3·9–6·0]), and previous admission with cardiac (1·6 [1·3–2·0]) or respiratory disease (1·5 [1·1–2·0]). Bile-duct exploration and acute pancreatitis—although risk factors on univariate analysis (2·8 [2·2–3·7] and 3·0 [2·0–4·6], respectively)—were not independent risk factors in the model with the above variables (1·1 [0·8–1·5] and 1·21 [0·8–1·5], respectively).

With 1981–82 as the reference category, risk-adjusted 90-day postoperative mortality fell during the prelaparoscopic decade (table 2 and figure, F). With 1989–90 as the reference category there was no signicant change in mortality during the laparoscopic period (1990–98). 1-year mortality (excluding 90-day mortality) did not change between 1981 and 1998 (data not shown).

Comparison of laparoscopic and open cholecystectomies (1996–98)

Table 3 shows that laparoscopic patients were younger, more likely to be female, and to have an elective admission, less likely to require ERCP, bile-duct exploration, or readmission within 28 days, and less likely to die by 90 days postoperatively. After adjustment for case mix, hospital stay was 3·1 days shorter.

Caseload and laparoscopic learning curve

The number of surgeons doing cholecystectomy increased by 11% from a median of 151 in the 1980s to 169 in 1998, reflecting a general increase in the number of surgeons. However, there was no significant change in the number of cholecystectomies done each year per surgeon (median 23 [IQR 10–36]). In the 1990s, there was a widening in the caseload range compared with the 1980s: the proportion of surgeons doing less than 15 per year increased from 30% in 1980s to 35% in 1993–98 (difference 5% [95% CI 1–8]), whereas the proportion doing more than 40 per year also increased from 16 to 21% (difference 5% [2–8]).

Table 4 shows the effect of cumulative number of laparoscopic cholecystectomies done by a surgeon and the effect of annual surgeon caseload. Procedures done early in the cumulative experience of an individual surgeon were associated with a significantly longer postoperative stay and, for the first ten procedures, significantly increased postoperative mortality. A smaller annual cholecystectomy workload was associated with lower odds of a procedure being laparoscopic, a significantly longer postoperative stay, and, for those doing fewer than three procedures a year, a substantial increase in the odds of postoperative death. A similar trend for postoperative stay was seen when the analyses in table 4 were repeated for laparoscopic procedures alone. The odds ratios for postoperative mortality were even higher for surgeons' first ten laparoscopic procedures (2·33 [1·18–4·58], p=0·015) and operations by surgeons doing one or two cholecystectomies a year (5·13 [2·29–11·51], p<0·0001). Although there was a trend to higher mortality in the 101–200 procedure group, when the analysis was confined to laparoscopic procedures, the odds ratio was not significantly higher (1·30 [0·58–2·88]).

Section snippets

Discussion

The strength of this study is that it uses a large population-based dataset from a defined geographical area. The record-linkage database had rigorous quality controls and it is likely that most procedures done in Scotland were included. Our study had some limitations, including the lack of information about postoperative morbidity. Comparisons between open and laparoscopic procedures are limited by selection bias, which may occur both in the choice of low-risk patients for the laparoscopic

References (29)

  • R Orlando et al.

    Laparoscopic cholecystectomy: a statewide experience

    Arch Surg

    (1993)
  • CA Steiner et al.

    Surgical rates and operative mortality for open and laparoscopic cholecystectomy in Maryland

    N Engl J Med

    (1996)
  • AY Chen et al.

    Growing use of laparoscopic cholecystectomy in the national Veterans Affairs Surgical Risk Study: effects on volume, patient selection, and selected outcomes

    Ann Surg

    (1998)
  • AJ McMahon et al.

    Bile duct injury and bile leakage in laparoscopic cholecystectomy

    Br J Surg

    (1995)
  • Cited by (128)

    • Evolution of bile duct repair in a low and middle-income country (LMIC): a comparison of diagnosis, referral, management and outcomes in repair of bile duct injury after laparoscopic cholecystectomy from 1991 to 2004 and 2005–2017

      2020, HPB
      Citation Excerpt :

      Several important changes regarding LC occurred in the early 2000's which underpin the rationale for dividing our cohort of patients into the two time periods from 1991 to 2004 and from 2005 to 2017. Population-based studies from Europe, the USA and Australia have shown that the proportion of LCs performed annually increased dramatically in the first 10 years after it was introduced.17–20 Interestingly, in these studies the number of LCs continued to rise, significantly increasing the total number of cholecystectomies performed annually beyond pre-LC open cholecystectomy rates.17

    • Chapter 17 - Role of nuclear medicine in diagnosis and management of hepatopancreatobiliary disease

      2017, Blumgart's Surgery of the Liver, Biliary Tract and Pancreas: Sixth Edition
    • Open versus laparoscopic cholecystectomy in acute cholecystitis. Systematic review and meta-analysis

      2015, International Journal of Surgery
      Citation Excerpt :

      The outcome was investigated either evaluating all the included studies, or evaluating only the randomized trials. Although cholecystectomy has a relatively low operative mortality of 0.4–0.6% [36,37] post-operative mortality is associated with emergency admission, co-morbid cardiorespiratory disease, and advanced age [38]. The present analysis showed the positive impact on mortality of LC in AC.

    View all citing articles on Scopus
    View full text