Efficacy and Safety of Granisetron 0.1 mg for Postoperative Nausea and Vomiting (PONV): A Quantitative Systematic Review

Piotr K. Janicki, MD, PhD
Department of Anesthesiology
Pennsylvania State University College of Medicine
Milton S. Hershey Medical Center
Michael Coyne, RPh, MS
Associate Vice President and Director of Pharmacy
Staten Island University Hospital

SUMMARY

Purpose: Our objective was to compare the efficacy and safety of granisetron 0.1 mg versus granisetron ≥ 1.0 mg for prophylaxis of postoperative nausea and vomiting (PONV).
Summary: MEDLINE and Cochrane Library databases were searched for published randomized clinical trials (RCTs) with granisetron 0.1 or ≥ 1.0 mg for PONV prophylaxis. Ten RCTs met inclusion criteria. The relative risk (RR), and number needed to treat (NNT) to prevent one additional event following were calculated for efficacy and data obtained from each pairwise comparison. Ten RCTs met inclusion criteria. During the early postoperative period, the NNTpool value for granisetron 0.1 mg was superior to placebo for absence of vomiting, whereas the NNTpool value for granisetron 1.0 mg did not significantly differ from that for placebo, but may have included confounding factors. Granisetron 0.1 and 1.0 mg had NNTpool values for absence of nausea that were significantly greater than placebo and overlapped, although granisetron 0.1 mg may have had confounding factors. The combined sets of granisetron 0.1/1.0 mg were homogeneous and did not significantly differ from granisetron doses > 1.0 mg for absence of vomiting and nausea. During the overall postoperative period, the NNTpool values for granisetron 0.1 and 1.0 mg overlapped and were superior to placebo for absence of vomiting and absence of nausea. Granisetron 1.0 mg had a superior NNTpool value versus placebo for need for rescue medication, and although data were insufficient for granisetron 0.1 mg comparison, the combined set was superior to placebo. For incidence of adverse effects, the combined set granisetron 0.1/1.0 mg was homogeneous, implying no apparent difference between granisetron 0.1 and 1.0 mg comparisons with placebo.
Conclusions: Granisetron 0.1 mg appears to be as effective and safe as granisetron 1.0 mg for PONV during the early and overall postoperative periods.

INTRODUCTION

Postoperative nausea and vomiting (PONV) are common complications following surgery. Patients fear PONV more than postoperative pain (Macario et al, 1999). Consequences of PONV range from patient discomfort and dissatisfaction to clinically significant events such as wound dehiscence and bleeding, dehydration and electrolyte imbalance, and aspiration pneumonitis. Unplanned patient admissions after outpatient surgery have occurred in up to 2% of patients due to PONV (Fortney et al, 1998).
Studies and meta-analysis of PONV show high variability in PONV rates (Cohen et al, 1994; Visser et al, 2001; Wu et al, 2002). Overall incidences of postoperative vomiting range from 12%–26% (Cohen et al, 1994) , while the overall incidence of postoperative nausea ranges from 22%–38% (Cohen et al, 1994). Although newer pharmacologic modalities such as the 5HT3 receptor antagonists (5HT3 RAs) have been introduced in the past 10 years, 25%–30% of patients continue to experience PONV within 24 hours after surgery (Cohen et al, 1994).
There are three 5HT3 RAs approved by the US Food and Drug Administration (FDA) for the prevention of PONV: granisetron, ondansetron, and dolasetron. The current FDA-approved dose of granisetron for this indication is 1.0 mg. The current consensus guideline dose for granisetron is 0.35–1.0 mg to be administered at the end of surgery (Gan et al, 2003). Our objective was to determine and to compare the efficacy and safety of low-dose granisetron, i.e., 0.1 mg, with granisetron 1.0 mg for prevention of PONV.

METHODS AND AIMS

Search Strategy and Data Extraction

A MEDLINE search for all randomized controlled trials (RCTs), published in any language from 1966 through April 1, 2006, was conducted. Included studies were classified as human, included the terms “granisetron” or “Kytril,” and excluded the terms “cancer,” “chemotherapy,” “chemotherapeutic,” “cisplatin,” “carboplatin,” “stem cell,” “pediatric,” or “children” in the title. An additional search was conducted of the Cochrane Library for relevant RCTs. Studies of pediatric populations (< 18 years of age) and pertaining to indications other than PONV and without at least one treatment arm with granisetron 0.1 or 1.0 mg, alone or in combination with another antiemetic agent, were excluded.
Ten RCTs were included for analysis, all but one (Gan et al, 2005) with three or more treatment arms. The 10 RCTs had a total of 29 randomized pairwise treatment comparisons (Npair): 9 between granisetron 0.1 mg and placebo (Npair = 3) or active agent (Npair = 6) and 20 between granisetron 1.0 mg and placebo (Npair = 7) or active agents (Npair = 13). Within pairwise comparisons, granisetron 0.1 mg (alone or in combination) was designated as “intervention,” with the comparison (placebo or active) as “control.” Pairwise comparisons with granisetron 1.0 mg (alone or in combination) were designated similarly, unless the comparator was granisetron 0.1 mg.
All data extracted from each granisetron 0.1 mg– or 1.0 mg–based treatment arm were dichotomous. Efficacy measures were the proportion of patients who did not vomit, had no nausea, and required rescue medication. Efficacy measures were selected as is (e.g., rather than the proportion of patients exhibiting vomiting), as they were the most commonly reported and did not require knowledge of the number and timing of patient withdrawals. Outcome data were categorized into postoperative periods previously described (Habib and Gan, 2004): early (0–6 hours), late (6–24 hours), and overall (0–24 hours). All available safety data were extracted.

Within-Study Statistical Analysis

For each pairwise comparison, we calculated the relative risk (RR, i.e., the number of events divided by the total number of patients in the group = experimental event rate [EER] divided by the risk of an event in the control group [control event rate or CER]) and the number needed to treat (NNT) to prevent one additional event, which is the inverse of the absolute risk reduction. For each outcome, the 95% confidence interval (95% CI) was calculated: those crossing 1.0 for RR were considered not significant; those including positive and negative values for NNT were considered not significant. Significant RR values > 1.0 implied that the measured proportion in the intervention group (i.e., 0.1 or 1.0 mg granisetron) of the pairwise comparison was significantly greater than that of the control group; significant RR and NNT values < 1.0 implied that the measured proportion in the intervention group was significantly less than for the control group. Significant NNT values favoring the control group were defined as number needed to harm (NNH). As per convention, NNT values were rounded up and NNH values rounded down to the nearest integer.
For each adverse event (AE) reported in a study, chi-square (?2) with significance set at P < 0.05 with 2-tailed probability was used to test for potential differences across study treatment groups. The Yates’ correction for continuity in 2 * 2 tables was used when appropriate. If ?2 testing of a study was significant, pairwise comparisons were conducted with 2 * 2 ?2 testing and the Bonferroni adjustment (0.05/number of pairwise comparison within a study) to identify significant interactions.

Meta-Analyses (Across-Studies) Statistical Analysis

Meta-analyses were performed only if the data set included at least three pairwise comparisons (Npair ≥ 3) obtained from at least three independent studies. Pairwise comparisons were pooled to estimate effect sizes (ES) of granisetron 0.1 and 1.0 mg for efficacy measures during each postoperative period and for reported safety outcomes. ES were estimated from pooling RRs (RRpool) and NNTpool or NNHpool along with pooled 95% CIs. The random effects model, which assumes random variations both within and among studies, was used for RRpool and NNTpool or NNHpool. Statistical significance for pooled ES values was determined in the same manner as for nonpooled ES. Within each postoperative period, if there were insufficient data for meta-analysis of direct comparison between granisetron 0.1 and 1.0 mg, whenever possible, meta-analyses were conducted in which these agents were compared with common comparators, i.e., placebo, granisetron doses > 1.0 mg, and ondansetron (the validity of meta-analyses using indirect comparisons has been established by Song et al [2003]). Because of possible serialization by the Fujii et al group that might unduly affect meta-analyses (Apfel et al, 1999; Kranke et al, 1999; Kranke et al, 2000; Kranke et al, 2001), whenever possible, meta-analyses were conducted with and without such studies. Statistical calculations were performed using RevMan, version 4.2 (Cochrane Collaboration, 2003) and MedCalc, version 7.6.

RESULTS
Included Studies

A total of 110 RCTs were identified, 43 of which were excluded because they were not relevant to PONV and/or used a pediatric population, and 57 because they did not contain a treatment arm using granisetron 0.1 or 1.0 mg. Study populations, time of administration, and description of treatments for the 10 RCTs that met our inclusion criteria are listed in Table 1.

PDF of Tables

Granisetron 0.1 mg was administered in at least one treatment arm in four studies

(D'Angelo et al, 2005; Gan et al, 2005; Taylor et al, 1997; Wilson et al, 1996); granisetron 1.0 mg was administered in eight studies (Dua et al, 2004; Fujii et al, 1998; Fujii et al, 2001a; Fujii et al, 2001b; Hanaoka et al, 2004; Janknegt et al, 1999; Taylor et al, 1997; Wilson et al, 1996). Only two studies had treatment arms with both granisetron 0.1 and 1.0 mg (Taylor et al, 1997; Wilson et al, 1996).

Early Postoperative Period (0–6 Hours)
Within Studies

RR and NNT values for all pairwise comparison data for the early postoperative period (0–6 hours) are presented in Table 2.
Granisetron 0.1 mg vs. Granisetron 1.0 mg—Two studies provided direct pairwise comparisons between granisetron 0.1 and 1.0 mg during the early postoperative period. Taylor et al (1997) observed no significant difference in the absence of vomiting or nausea, whereas Wilson et al (1996) observed a significant difference favoring granisetron 1.0 mg over granisetron 0.1 mg for both absence of vomiting and absence of nausea. However, Taylor et al (1997) observed that the need for rescue medication favored granisetron 0.1 mg, whereas Wilson et al (1996) provided no data.
Granisetron vs. Placebo?Granisetron 0.1 mg was significantly superior to placebo in one of three pairwise comparisons for absence of vomiting (D'Angelo et al, 2005), in two of three for absence of nausea (D'Angelo et al, 2005; Taylor et al, 1997), and in both studies that provided data on rescue medication (D'Angelo et al, 2005; Taylor et al, 1997). Granisetron 1.0 mg was significantly superior to placebo in two of four pairwise comparisons for absence of vomiting (Taylor et al, 1997; Wilson et al, 1996) but was also significantly inferior in one pairwise comparison (Hanaoka et al, 2004). Granisetron 1.0 mg was significantly superior to placebo in two of three pairwise comparisons for absence of nausea (Taylor et al, 1997; Wilson et al, 1996).
Granisetron 0.1 or 1.0 mg vs. Granisetron Doses > 1.0 mg?Data for granisetron 0.1 mg versus granisetron doses > 1.0 mg during the early postoperative period were provided only by Taylor et al (Taylor et al, 1997), who reported no significant difference between granisetron 0.1 mg and granisetron 3.0 mg for absence of vomiting or nausea, or use of rescue medication. Four studies provided efficacy data for granisetron 1.0 mg versus granisetron 2.0, 3.0, and/or 4.0 mg (Fujii et al, 1998; Hanaoka et al, 2004; Taylor et al, 1997; Wilson et al, 1996). Granisetron 1.0 mg was significantly inferior to granisetron 4.0 mg for absence of vomiting in one study (Fujii et al, 1998) but did not significantly differ in four other pairwise comparisons for absence of vomiting or of nausea (Fujii et al, 1998; Hanaoka et al, 2004; Taylor et al, 1997; Wilson et al, 1996). However, granisetron 1.0 mg was significantly inferior to granisetron 3.0 mg for rescue medication use in the only pairwise comparison (Taylor et al, 1997).
Granisetron 0.1 or 1.0 mg vs. Ondansetron?Only Gan et al (2005) provided data for the early postoperative period (for granisetron 0.1 mg versus ondansetron, each in combination with dexamethasone). The granisetron 0.1 mg–based treatment arm did not significantly differ from ondansetron in absence of vomiting or nausea. None of the studies provided efficacy data comparing granisetron 1.0 mg versus ondansetron.

Meta-analysis (0–6 Hours)

RRpool and NNTpool values for sets of pairwise comparison data during the early postoperative period that met inclusion criteria are presented in Table 3.
Granisetron vs. Placebo?The pooled pairwise comparison of granisetron 0.1 mg versus placebo for absence of vomiting and absence of nausea during the early postoperative period significantly favored granisetron 0.1 mg. Similarly, granisetron 1.0 mg versus placebo and granisetron 0.1/1.0 mg versus placebo significantly favored granisetron therapy for both absence of vomiting and absence of nausea. There were insufficient data to assess granisetron 0.1 and/or 1.0 mg, each versus placebo for the need for rescue medication.
Granisetron 0.1 or 1.0 mg vs. Granisetron Doses > 1.0 mg?There were insufficient data for pooled pairwise comparisons of granisetron 0.1 mg versus granisetron doses > 1.0 mg for any efficacy measure. Pairwise comparisons of granisetron 1.0 mg versus granisetron doses > 1.0 mg and granisetron 0.1/1.0 mg versus granisetron > 1.0 mg for absence of vomiting and of nausea were all nonsignificant, regardless of whether Fujii et al studies were included.

Late Postoperative Period (6-24 Hours)
Within Studies

Few data were available during the late postoperative period (6–24 hours), and none on the absence of vomiting or nausea. Three studies provided data on the need for rescue medication (D'Angelo et al, 2005; Dua et al, 2004; Gan et al, 2005). There was no significant difference in rescue medication use between granisetron 0.1 mg and dexamethasone versus ondansetron and dexamethasone (Gan et al, 2005), granisetron 0.1 mg versus placebo or granisetron 0.3 mg (D'Angelo et al, 2005), or between granisetron 1.0 mg and placebo or ondansetron (Dua et al, 2004).
Meta-analysis (6–24 Hours)
There were insufficient data to pool pairwise comparisons of granisetron 0.1, 1.0, or 0.1/1.0 mg with placebo or granisetron doses > 1.0 mg for any efficacy measure.

Overall Postoperative Period Within Studies

RR and NNT values for all pairwise comparison data for the overall postoperative period (0–24 hours) are presented in Table 4.
Granisetron 0.1 mg vs. Granisetron 1.0 mg—Only Taylor et al (1997) and Wilson et al (1996) provided direct pairwise comparisons between granisetron 0.1 and 1.0 mg during the overall postoperative period. Taylor et al (1997) observed nonsignificant NNT values for all efficacy measures. Wilson et al (1996), however, observed significant NNT values favoring granisetron 1.0 mg over 0.1 mg for absence of both vomiting and nausea.
Granisetron vs. Placebo—Three studies (D'Angelo et al, 2005; Taylor et al, 1997; Wilson et al, 1996) provided pairwise comparisons of granisetron 0.1 mg versus placebo during the overall postoperative period. D’Angelo et al (2005) observed nonsignificant NNTs for all efficacy measures of granisetron 0.1 mg versus placebo, whereas Taylor et al (1997) observed that granisetron 0.1 mg was superior for all efficacy measures. Wilson et al (1996) reported that granisetron 0.1 mg was superior to placebo for absence of vomiting but not for absence of nausea. Among the seven studies providing pairwise comparisons of granisetron 1.0 mg versus placebo (Dua et al, 2004; Fujii et al, 1998; Fujii et al, 2001a; Fujii et al, 2001b; Hanaoka et al, 2004; Taylor et al, 1997; Wilson et al, 1996), Fujii et al (1998, 2001a, 2001b) and Dua et al (2004) reported no significant differences in any efficacy measure. The remaining studies yielded NNTs favoring granisetron 1.0 mg over placebo for absence of vomiting (Hanaoka et al, 2004; Taylor et al, 1997; Wilson et al, 1996), absence of nausea (Taylor et al, 1997; Wilson et al, 1996), and need for rescue medication (Hanaoka et al, 2004; Taylor et al, 1997).
Granisetron 0.1 or 1.0 mg vs. Granisetron Doses > 1.0 mg?Two studies (Taylor et al, 1997; Wilson et al, 1996) provided data on granisetron 0.1 mg versus granisetron doses > 1.0 mg. Taylor et al (1997) observed no significant NNTs for any efficacy measure of granisetron 0.1 mg versus granisetron 3.0 mg. Wilson et al (1996), on the other hand, reported that granisetron 0.1 mg appeared inferior to granisetron 3.0 mg for absence of vomiting and absence of nausea.
Granisetron 0.1 or 1.0 mg vs. Ondansetron?Gan et al (2005) had nonsignificant NNTs for granisetron 0.1 mg versus ondansetron (each in combination with dexamethasone) for absence of vomiting or nausea. Similarly, Dua et al (2004) had no significant NNT for any efficacy measure.
Meta-analysis (0–24 Hours)
RRpool and NNTpool values for sets of pairwise comparison data for the overall postoperative period are presented in Table 5.
Granisetron vs. Placebo?Pooled pairwise comparisons of granisetron 0.1 mg, granisetron 1.0 mg (with or without Fujii et al studies), and the combined set of granisetron 0.1/1.0 mg (without Fujii et al studies), each versus placebo, showed that treatment with granisetron was significantly superior to placebo for absence of vomiting and absence of nausea. The confidence intervals for NNTpool values for granisetron 0.1 mg, for absence of both vomiting and nausea, overlapped that for granisetron 1.0 mg. There were insufficient data for pooling pairwise comparisons of granisetron 0.1 mg versus placebo for rescue medication. However, the pairwise set of granisetron 0.1/1.0 mg versus placebo significantly favored granisetron therapy regardless of whether Fujii et al comparisons were included. Pairwise comparison of granisetron 1.0 mg versus placebo, with or without Fujii et al studies, significantly favored granisetron for rescue medication.
Granisetron 0.1 or 1.0 mg vs. Granisetron Doses > 1.0 mg?There were insufficient data for pooled pairwise comparisons of granisetron 0.1 mg versus granisetron doses > 1.0 mg for any efficacy measure during the overall postoperative period. NNTpool values for pooled pairwise comparisons of granisetron 1.0 mg versus granisetron doses > 1.0 mg were not significant for absence of vomiting or nausea. Granisetron > 1.0 mg appeared to be superior to granisetron 0.1/1.0 mg for rescue medication use, but the set appeared to be heterogeneous (P < 0.0001). There was no significant effect between granisetron 0.1/1.0 mg and granisetron doses > 1.0 mg for absence of vomiting, although treatment significantly favored granisetron > 1.0 mg for absence of nausea and use of rescue medication.

SAFETY
Within-Study Analysis

None of the studies reported treatment-related deaths. Serious AEs were infrequently reported. Hanaoka et al (2004) reported ventricular fibrillation in one patient treated with granisetron 1.0 mg, but it was not considered related to study medication. More recently, Gan et al (2005) reported four serious AEs among patients treated with ondansetron 4.0 mg and dexamethasone 8.0 mg compared with two among patients treated with granisetron 0.1 mg and dexamethasone (P = 0.72). However, only 7 of 10 studies reported any quantitative data on safety (D'Angelo et al, 2005; Dua et al, 2004; Fujii et al, 1998; Gan et al, 2005; Hanaoka et al, 2004; Janknegt et al, 1999; Taylor et al, 1997). The proportion of patients with AEs was reported in four studies, none of which had significant differences within treatment groups (D'Angelo et al, 2005; Fujii et al, 1998; Gan et al, 2005; Taylor et al, 1997). Only two studies provided the proportion of treatment-related AEs (Gan et al, 2005; Hanaoka et al, 2004). No significant differences among treatment groups were observed (Gan et al, 2005; Hanaoka et al, 2004).
AEs reported in only one or two studies included abdominal pain (Taylor et al, 1997), anemia (D'Angelo et al, 2005; Taylor et al, 1997), anxiety (Dua et al, 2004; Taylor et al, 1997), bradycardia (Gan et al, 2005; Taylor et al, 1997), constipation (Fujii et al, 1998; Taylor et al, 1997), diarrhea (D'Angelo et al, 2005), drowsiness (Dua et al, 2004), dry mouth (Janknegt et al, 1999), extrapyramidal symptoms (Janknegt et al, 1999), flatulence (D'Angelo et al, 2005), hepatic enzyme increase (Taylor et al, 1997), hypertension (Taylor et al, 1997), insomnia (Dua et al, 2004), oliguria (D'Angelo et al, 2005), pain (Taylor et al, 1997), pyrexia/fever (Hanaoka et al, 2004), sedation (Janknegt et al, 1999), and skin discoloration (Janknegt et al, 1999). The only significant difference among treatment groups in any of these studies was an increase in hepatic enzymes (?2 = 9.54, P = 0.02) (Taylor et al, 1997).
AEs reported in at least three studies included dizziness (Dua et al, 2004; Fujii et al, 1998; Hanaoka et al, 2004; Janknegt et al, 1999), which was not reported in any group of patients administered granisetron 0.1 mg, pruritis (D'Angelo et al, 2005; Gan et al, 2005; Hanaoka et al, 2004); and most frequently, headache (D'Angelo et al, 2005; Dua et al, 2004; Fujii et al, 1998; Gan et al, 2005; Hanaoka et al, 2004; Janknegt et al, 1999; Taylor et al, 1997). No significant differences among treatment groups were observed in studies in which patients reported dizziness (Dua et al, 2004; Fujii et al, 1998; Hanaoka et al, 2004; Janknegt et al, 1999) or pruritis (D'Angelo et al, 2005; Gan et al, 2005; Hanaoka et al, 2004). In the only study that provided safety data for a pairwise comparison between granisetron 0.1 mg and 1.0 mg (Taylor et al, 1997), there were no significant NNTs for the incidence of any AE or headaches.

Meta-analysis of Safety

The RRpool and NNHpool values for sets of pairwise comparison data for the safety measures that met criteria for meta-analysis (i.e., any AE, dizziness, headaches, and pruritis) are presented in Table 6.
Any AE—The set of all pairwise comparisons of granisetron 0.1/1.0 mg (Q = 3.46, P = 0.99) and the set of granisetron 0.1 mg pairwise comparisons (Q = 2.59, P = 0.76) that provided data on the incidence of AEs appeared to be homogeneous and to favor granisetron 0.1 mg. The set of granisetron 1.0 mg pairwise comparisons was homogeneous (Q = 0.22, P > 0.99) but did not significantly favor granisetron 1.0 mg or comparators. Although there were insufficient data from pairwise comparisons of granisetron 0.1 mg versus placebo for meta-analysis, when this set was combined with the set of pairwise comparison of granisetron 1.0 mg versus placebo (which was not significant), the resulting set was homogeneous (Q = 2.00, P = 0.74) and significantly favored therapy with either granisetron 0.1 or 1.0 mg versus placebo, meaning that for every 13 patients treated with granisetron 0.1 or 1.0 mg, one patient would be prevented from developing any AE. Insufficient data were available for meta-analysis of granisetron 0.1/1.0 mg versus granisetron doses > 1.0 mg for any AE. However, the set of pairwise comparisons of granisetron 0.1 mg versus active agents (ondansetron with dexamethasone, and granisetron at 0.3, 1.0, and 3.0 mg) was homogeneous (Q = 0.52, P = 0.92), as was the set of pairwise comparisons of granisetron 1.0 mg versus active agents (ondansetron, and granisetron 2.0, 3.0, and 4.0 mg) (Q = 0.52, P = 0.92) and granisetron 0.1/1.0 mg versus active agents, whether including (Q = 0.98, P > 0.99) or excluding Fujii et al (Q = 0.90, P = 0.97). The NNHpool value was not significant for any of these pairwise sets.
Headaches—The sets of all pairwise comparisons of granisetron 0.1 mg (Q = 5.19, P = 0.39), granisetron 1.0 mg (Q = 15.21, P = 0.17), and granisetron 0.1/1.0 mg that provided data on the incidence of headaches (Q = 25.27, P = 0.09) were homogeneous. None significantly differed from comparators. Although there were insufficient data for meta-analysis of pairwise comparisons of granisetron 0.1 mg versus placebo, the set of granisetron 0.1/1.0 mg versus placebo, with (Q = 4.70, P = 0.45) or without Fujii et al studies (Q = 4.52, P = 0.34), was homogeneous but did not yield a significant NNH. Insufficient data were available for meta-analysis of granisetron 0.1/1.0 mg versus granisetron doses > 1.0 mg for headaches. The sets of pairwise comparisons of granisetron 1.0 mg (Q = 2.62, P = 0.45) and granisetron 0.1/1.0 mg, each versus granisetron doses > 1.0 mg, were homogeneous (Q = 3.13, P = 0.54) but did not yield significant NNHpool values. Pairwise comparisons of granisetron 0.1 mg (Q = 3.07, P = 0.38) versus active agents (ondansetron with dexamethasone, and granisetron at 0.3, 1.0, and 3.0 mg), granisetron 1.0 mg versus active agents (ondansetron, droperidol, and granisetron 2.0, 3.0, and 4.0 mg) with (Q = 0.11, P = 1.00) or without Fujii et al studies (Q = 7.21, P = 0.21), and granisetron 0.1/1.0 mg (Q = 15.23, P = 0.17) were homogeneous, but none yielded a significant NNHpool value.
Pruritis—There were insufficient data to assess pairwise comparisons of granisetron 0.1 or 1.0 mg versus any comparator for incidence of pruritis. The set of pairwise comparisons of granisetron 0.1/1.0 mg was homogeneous (Q = 1.10, P = 0.89) but with a nonsignificant NNHpool value. Granisetron 0.1/1.0 mg versus active comparators (ondansetron, droperidol, and granisetron 3.0 mg) was homogeneous (Q = 0.47, P = 0.79) but yielded a nonsignificant NNHpool value.
Dizziness—The incidence of dizziness was reported only in studies involving granisetron 1.0 mg. The complete set of granisetron 1.0 mg versus comparators for incidence of dizziness was homogeneous but did not have a significant NNHpool value.

DISCUSSION

This meta-analysis was designed to directly compare the efficacy and safety of granisetron 0.1 mg, which has a considerably lower cost, with that of granisetron 1.0 mg for PONV based on published RCTs. However, because only two studies provided direct comparison data, we used established indirect meta-analytic comparison techniques (Song et al, 2003). Both efficacy and safety data for granisetron 0.1 and 1.0 mg were compared with those of mutual comparators, placebo and granisetron doses > 1.0 mg. Meta-analytic techniques were employed whenever at least three pairwise comparisons were available from at least three independent studies.
During the early postoperative period, the NNTpool value for granisetron 0.1 mg was homogeneous and superior to that of placebo for absence of vomiting, whereas the NNTpool value for granisetron 1.0 mg did not significantly differ from that of placebo but was heterogeneous, possibly indicating the presence of confounding factors. Both granisetron 0.1 mg and granisetron 1.0 mg had NNTpool values for absence of nausea that were significantly greater than placebo and overlapped, although granisetron 0.1 mg may have had confounding factors. The combined set of granisetron 0.1/1.0 mg was homogeneous, indicating the absence of confounding factors, and did not significantly differ from granisetron doses > 1.0 mg for absence of vomiting and nausea. Based on these findings, it appears that both dose levels of granisetron are similarly effective for PONV during the early postoperative period and statistically indistinguishable from higher dose levels of granisetron.
During the overall postoperative period, the NNTpool value for granisetron 0.1 and 1.0 mg were homogeneous, had overlapping CIs, and were superior to that of placebo for absence of vomiting and absence of nausea. Granisetron 1.0 mg had a superior NNTpool value versus placebo for need for rescue medication, and although there were insufficient data for granisetron 0.1 mg comparison, the combined set was superior to placebo and homogeneous. Because of insufficient granisetron 0.1 mg data, comparisons with granisetron doses > 1.0 mg were less clear; all sets were heterogeneous and may have had confounding factors. Granisetron 0.1/1.0 mg did not significantly differ from granisetron doses > 1.0 mg for absence of vomiting but was associated with an NNTpool value that favored granisetron > 1.0 mg for need for rescue medication. Outcomes for absence of nausea conflicted. Based on these findings, both dose levels also appear similarly effective for PONV during the overall postoperative period. Overall, these efficacy and safety findings accord with a retrospective cohort study of 400 patients, most of whom received granisetron 0.1 mg perioperatively for PONV (Shillington et al, 2005).
The efficacy and safety findings of dosing with granisetron 0.1 mg shown in the analysis are accompanied by an economic advantage. Treatment with granisetron 0.1 mg, compared with standard granisetron 1.0 mg, could reduce treatment costs by 90%, and by 30% compared with ondansetron 4.0 mg and dolasetron 12.5 mg.
As with any meta-analysis of studies, ours was limited by the use of study aggregate data as opposed to individual patient data, which are significantly more powerful. Other limitations included (1) deficiencies in reported efficacy and safety data; (2) insufficient studies for direct meta-analytic comparison of granisetron 0.1 mg versus granisetron 1.0 mg, necessitating indirect pairwise comparisons using treatment arms from RCTs with at least three treatment arms; and (3) insufficient treatment arms or studies from which to analyze heterogeneity, perform additional subset analyses, or address possible confounding factors. Additional limitations that must be considered when evaluating the results of this meta-analysis are the inclusion of RCTs in which: (1) granisetron was administered at different times (i.e., pre, intra-, or postoperatively) and (2) a second agent, such as dexamethasone, is included.

CONCLUSIONS

Granisetron 0.1 mg appears to be as effective as granisetron 1.0 mg for PONV during the early and overall postoperative periods and has a similar safety profile. The cost reduction associated with granisetron 0.1 mg should be considered when developing hospital guidelines for the prevention of PONV.

ACKNOWLEDGMENT

The authors wish to acknowledge Nelson Erlick, DPM, MS, for his statistical analyses and editorial assistance in the preparation of this manuscript.

REFERENCES

Apfel, C C, Kranke, P, Greim, C A, Roewer, N. (1999) Non-systematic Serial Publishing Is Not Appropriate and Ethically Questionable. Acta Anaesthesiologica Scandinavica. Vol. 43, 486-487.
Cochrane Collaboration. (2003) Review Manager (RevMan) [computer program]. Version 4.2 for Windows. Oxford, England.
Cohen, M M, Duncan, P G, DeBoer, D P, Tweed, W A. (1994) The Postoperative Interview: Assessing Risk Factors for Nausea and Vomiting. Anesthesia and Analgesia. Vol. 78, 7-16.
D'Angelo, R, Philip, B, Gan, T, Kovac, A, Hantler, C, Doblar, D, et al. (2005) A Randomized, Double-Blind, Dose-Ranging, Pilot Study of Intravenous Granisetron in Prevention of Post-Operative Nausea and Vomiting (PONV) in Patients Undergoing Abdominal Hysterectomy. European Journal of Anaesthesiology. Vol. 22, 774-779.
Dua, N, Bhatnagar, S, Mishra, S, Singhal, A K. (2004) Granisetron and Ondansetron for Prevention of Nausea and Vomiting in Patients Undergoing Modified Radical Mastectomy. Anaesthesia and Intensive Care. Vol. 32, 761-764.
Fortney, J T, Gan, T J, Graczyk, S, Wetchler, B, Melson, T, Khalil, S, et al; S3A-409 and S3A-410 Study Groups. (1998) A Comparison of the Efficacy, Safety, and Patient Satisfaction of Ondansetron Versus Droperidol as Antiemetics for Elective Outpatient Surgical Procedures. Anesthesia and Analgesia. Vol. 86, 731-738.
Fujii, Y, Tanaka, H, Kawasaki, T. (2001a) Preoperative Oral Granisetron for the Prevention of Postoperative Nausea and Vomiting After Breast Surgery. European Journal of Surgery. Vol. 167, 184-187.
Fujii, Y, Tanaka, H, Kawasaki, T. (2001b) Prophylaxis With Oral Granisetron for the Prevention of Nausea and Vomiting After Laparoscopic Cholecystectomy: A Prospective Randomized Study. Archives of Surgery. Vol. 136, 101-104.
Fujii, Y, Tanaka, H, Toyooka, H. (1998) Preoperative Oral Granisetron Prevents Postoperative Nausea and Vomiting. Acta Anaesthesiologica Scandinavica. Vol. 42, 653-657.
Gan, T J, Coop, A, Philip, B K. (2005) A Randomized, Double-Blind Study of Granisetron Plus Dexamethasone Versus Ondasetron Plus Dexamethasone to Prevent Postoperative Nausea and Vomiting in Patients Undergoing Abdominal Hysterectomy. Anesthesia and Analgesia. Vol. 101, 1323-1329.
Gan, T J, Meyer, T, Apfel, C C, Chung, F, Davis, P J, Eubanks, S, et al. (2003) Consensus Guidelines for Managing Postoperative Nausea and Vomiting. Anesthesia and Analgesia. Vol. 97, 62-71.
Habib, A S, Gan, T J. (2004) Evidence-Based Management of Postoperative Nausea and Vomiting: a Review. Canadian Journal of Anaesthesia. Vol. 51, 326-341.
Hanaoka, K, Toyooka, H, Kugimiya, T, Ohashi, Y. (2004) Efficacy of Prophylactic Intravenous Granisetron in Postoperative Emesis in Adults. Journal of Anesthesia. Vol. 18, 158-165.
Janknegt, R, Pinckaers, J W, Rohof, M H, Ausems, M E, Arbouw, M E, van, d, V, et al. (1999) Double-Blind Comparative Study of Droperidol, Granisetron and Granisetron Plus Dexamethasone as Prophylactic Anti-emetic Therapy in Patients Undergoing Abdominal, Gynaecological, Breast or Otolaryngological Surgery. Anaesthesia. Vol. 54, 1059-1068.
Kranke, P, Apfel, C C, Eberhart, L H, Georgieff, M, Roewer, N. (2001) The Influence of a Dominating Centre on a Quantitative Systematic Review of Granisetron for Preventing Postoperative Nausea and Vomiting. Acta Anaesthesiologica Scandinavica. Vol. 45, 659-670.
Kranke, P, Apfel, C C, Greim, C A, Roewer, N. (1999) Methodological Problems Arising From 'Serial Publishing' on the Effectiveness of Granisetron in PONV. British Journal of Anaesthesia. Vol. 82, 481-483.
Kranke, P, Apfel, C C, Roewer, N, Fujii, Y. (2000) Reported Data on Granisetron and Postoperative Nausea and Vomiting by Fujii et al. Are Incredibly Nice! Anesthesia and Analgesia. Vol. 90, 1004-1007.
Macario, A, Weinger, M, Carney, S, Kim, A. (1999) Which Clinical Anesthesia Outcomes Are Important to Avoid? The Perspective of Patients. Anesthesia and Analgesia. Vol. 89, 652-658.
Shillington, A, Farley, P A, Varkey, J A, Dempsey, C, Coop, A, Harshaw, Q, et al. (2005) Retrospective Cohort Study of Granisetron Administration for PONV Prophylaxis and Treatment. Hospital Pharmacy. Vol. 40, 592-598.
Song, F, Altman, D G, Glenny, A M, Deeks, J J. (2003) Validity of Indirect Comparison for Estimating Efficacy of Competing Interventions: Empirical Evidence From Published Meta-analyses. British Medical Journal. Vol. 326, 472.
Taylor, A M, Rosen, M, Diemunsch, P A, Thorin, D, Houweling, P L. (1997) A Double-Blind, Parallel-Group, Placebo-Controlled, Dose-Ranging, Multicenter Study of Intravenous Granisetron in the Treatment of Postoperative Nausea and Vomiting in Patients Undergoing Surgery With General Anesthesia. Journal of Clinical Anesthesia. Vol. 9, 658-663.
Visser, K, Hassink, E A, Bonsel, G J, Moen, J, Kalkman, C J. (2001) Randomized Controlled Trial of Total Intravenous Anesthesia With Propofol Versus Inhalation Anesthesia With Isoflurane-Nitrous Oxide: Postoperative Nausea With Vomiting and Economic Analysis. Anesthesiology. Vol. 95, 616-626.
Wilson, A J, Diemunsch, P, Lindeque, B G, Scheinin, H, Helbo-Hansen, H S, Kroeks, M V, et al. (1996) Single-Dose I.V. Granisetron in the Prevention of Postoperative Nausea and Vomiting. British Journal of Anaesthesia. Vol. 76, 515-518.
Wu, C L, Berenholtz, S M, Pronovost, P J, Fleisher, L A. (2002) Systematic Review and Analysis of Postdischarge Symptoms After Outpatient Surgery. Anesthesiology. Vol. 96, 994-1003.

Home • Journals • Search • Rules for Authors • Submit a Paper • Sponsor us