Association of Initial Treatment With Antibiotics vs Surgery With Treatment Success and Disability in Subgroups of Children With Uncomplicated Appendicitis

Abstract

This study investigates differences in 1-year outcomes by clinical and socioeconomic factors among US children with uncomplicated appendicitis who were treated with antibiotics vs surgery.

Initial treatment with antibiotics alone is a reasonable option for children presenting with uncomplicated appendicitis, as demonstrated in a patient choice study that found nonoperative management was successful in 67% at 1 year and led to fewer disability days compared with laparoscopic appendectomy.¹ Because differences in outcomes by age, race and ethnicity, family income, and transfer status have been described in patients undergoing appendectomy, we evaluated whether the primary outcomes of that study varied in subgroups based on clinical and socioeconomic factors.^2,3

Methods

This was a planned secondary analysis of a multi-institutional, nonrandomized, controlled interventional study investigating nonoperative management and surgery for children with uncomplicated appendicitis. Caregivers of children aged 7 to 17 years treated at 10 US children’s hospitals from May 2015 to October 2018 chose treatment with either antibiotics alone or urgent appendectomy. Follow-up was through October 2019. The primary results have been previously published, and the study protocol is available in Supplement 1.¹ Comparisons between groups used propensity score methods to account for confounding by baseline characteristics by balancing covariates between treatment groups in the final analysis.^4,5,6 Multiple imputation was used to account for missing data in estimating the propensity score. Analyses used inverse probability of treatment weighting, as estimated with the propensity score, to calculate adjusted treatment estimates for 2 primary outcomes, number of disability days experienced by children and success rate of nonoperative management at 1 year. Three factors were specified a priori for subgroup analyses: age (≤10 years or >10 years), annual household income (<$50 000 or ≥$50 000), and transfer status. Planned exploratory analyses included stratification by race and ethnicity, with sex, insurance status, primary caregiver education, and language spoken at home added post hoc. Number of household residents was not explored because of inconsistent reporting. Subgroup analyses used regression models weighted using inverse probability of treatment weighting to compare success rates of nonoperative management between subgroups, while tests of interactions were used for disability days. Statistical significance for the 6 comparisons was defined at the 2% level (for an overall family-wise error rate of 12%). Data management was performed in SAS version 9.4 (SAS Institute Inc) and analyses in Stata version 15.1 (StataCorp). Institutional review board approval was obtained at each site, and written informed consent was obtained from the caregivers of all participants.

Results

Of 1068 enrolled patients, 370 had nonoperative management and 698 had appendectomy. Complete follow-up at 1 year was available for 77% (284/370) of nonoperative management patients and 75% (522/698) of surgery patients. The baseline characteristics for each group were previously published.¹ Nonoperative management was successful at 1 year for 63.8% of patients aged 10 years or younger vs 68.1% of patients older than 10 years (P = .47), 65.9% of patients with household incomes of less than $50 000 vs 68.3% with household incomes of $50 000 or greater (P = .70), and 68.3% of patients presenting directly to the treating hospital vs 65.6% of transferred patients (P = .60). No significant differences in disability days at 1 year were identified between those who chose nonoperative management vs surgery by age (≤10 years: nonoperative management, 6.3 days, vs surgery, 10.3 days; difference, −4.0 [95% CI, −6.8 to −1.2] days; >10 years: nonoperative management, 6.6 days, vs surgery, 11.0 days; difference, −4.4 [95% CI, −6.0 to −2.7] days; P = .82); by income (<$50 000: nonoperative management, 7.1 days, vs surgery, 10.9 days; difference, −3.8 [95% CI, 6.2 to −1.4] days; ≥$50 000: nonoperative management, 5.8 days, vs surgery, 9.6 days; difference, −3.8 [95% CI, −5.4 to −2.2] days; P = .99); or by transfer status (not transferred: nonoperative management, 6.8 days, vs surgery, 11.3 days; difference, −4.4 [95% CI, −6.3 to −2.6] days; transferred: nonoperative management, 6.3 days, vs surgery, 10.3 days; difference, −4.1 [95% CI, −6.3 to −1.9] days; P = .87). There were no significant differences between nonoperative management and surgery based on sex, race and ethnicity, insurance status, primary caregiver education level, or primary language spoken at home (Table 1 and Table 2).

Table 1. Stratified IPTW-Adjusted Treatment Estimates for the Primary Outcome of 1-Year Success Rate of Nonoperative Management of Appendicitis.

Factors	No. of patients with medical record review	Success rate of nonoperative management at 1 y, No. (%) [95% CI]	P valuea
Patient age, y
≤10	93	59 (63.8) [52.7-73.5]	.47
>10	277	189 (68.1) [62.0-73.7]
Annual household income, $
<50 000	109	72 (65.9) [55.7-74.9]	.70
≥50 000	217	148 (68.3) [61.3-74.5]
Transfer status
Not transferred	211	144 (68.3) [61.3-74.6]	.60
Transferred	159	104 (65.6) [57.1-73.1]
Sexb
Male	229	155 (67.8) [60.9-74.0]	.72
Female	141	93 (65.9) [57.3-73.6]
Race and ethnicityb,c
Non-Hispanic White	199	140 (70.4) [63.2-76.8]	.14
Other race/ethnicity	171	107 (62.7) [54.6-70.1]
Insurance statusb
Private insurance	251	176 (70.0) [63.6-75.7]	.20
Medicaid	109	68 (62.5) [52.2-71.7]
Primary caregiver educationb
Some college or less	149	105 (70.2) [61.8-77.5]	.42
At least college degree	190	125 (65.8) [58.3-72.6]
Primary language spoken at homeb
English	284	191 (67.1) [61.0-72.6]	.82
Other	69	47 (68.6) [56.1-78.9]

^aMain effects for 1-year success rate of nonoperative management were tested through inverse probability of treatment weight (IPTW) logistic regression models. In all models, inference relied on robust sandwich-type standard errors to account for the estimated IPTWs.

^bFactor was added to the analysis post hoc.

^cRace and ethnicity were assessed because of evidence from the available literature suggesting an association with outcomes in pediatric patients treated for appendicitis. Related data were reported by caregivers using surveys with predefined categories listed in the table. Other includes Black, Hispanic, Asian, American Indian, Alaska Native, and biracial patients. For patients managed nonoperatively, this included 23 non-Hispanic Black patients, 37 Hispanic patients, 106 patients who reported another race or ethnicity, and 5 patients who did not report race or ethnicity.

Table 2. Stratified IPTW-Adjusted Treatment Estimates for the Primary Outcome of Disability Days at 1 Year.

Factors	No. of completers	Disability days at 1 y, mean (95% CI)		Difference (95% CI)	P value for interactiona
		Nonoperative management	Surgery
Patient age, y
≤10	161	6.34 (4.9-8.19)	10.33 (8.26-12.39)	−3.99 (−6.76 to −1.21)	.82
>10	645	6.64 (5.42-7.86)	11.00 (9.91-12.09)	−4.36 (−6.00 to −2.72)
Annual household income, $
<50 000	207	7.10 (5.46-8.74)	10.86 (9.09-12.64)	−3.77 (−6.18 to −1.35)	.99
≥50 000	504	5.84 (4.64-7.04)	9.60 (8.52-10.68)	−3.76 (−5.37 to −2.15)
Transfer status
Not transferred	462	6.81 (5.59-8.03)	11.26 (9.88-12.63)	−4.44 (−6.28 to −2.60)	.81
Transferred	344	6.26 (4.48-8.03)	10.34 (9.04-11.64)	−4.08 (−6.28 to −1.88)
Sexb
Male	506	6.44 (5.23-7.65)	10.64 (9.42-11.86)	−4.20 (−5.92 to −2.48)	.87
Female	300	6.81 (4.89-8.72)	11.26 (9.67-12.86)	−4.46 (−6.95 to −1.96)
Race and ethnicityb,c
Non-Hispanic White	484	6.37 (4.89-7.86)	10.11 (9.04-11.17)	−3.73 (−5.56 to −1.91)	.34
Other race/ethnicity	322	6.87 (5.55-8.20)	12.01 (10.20-13.81)	−5.14 (−7.38 to −2.89)
Insurance statusb
Private insurance	585	6.08 (4.99-7.27)	10.04 (9.06-11.02)	−3.96 (−5.50 to −2.42)	.38
Medicaid	198	7.89 (5.63-10.15)	13.55 (10.92-16.18)	−5.66 (−9.13 to −2.20)
Primary caregiver educationb
Some college or less	341	6.12 (4.65-7.60)	11.12 (9.63-12.62)	−5.00 (−7.10 to −2.90)	.11
At least college degree	389	6.67 (5.10-8.24)	9.32 (8.10-10.54)	−2.66 (−4.64 to −0.67)
Primary language spoken at homeb
English	676	6.51 (5.37-7.65)	10.20 (9.25-11.15)	−3.69 (−5.17 to −2.20)	.06
Other	92	6.01 (3.76-8.26)	14.44 (10.28-18.59)	−8.42 (−13.15 to −3.70)

^aInteraction effects between treatment choice and the stratification factor were estimated through inverse probability of treatment weight (IPTW) linear regression models for disability days with differences between treatment groups estimated through marginal standardization. In all models, inference relied on robust sandwich-type standard errors to account for the estimated IPTWs.

^bFactor was added to the analysis post hoc.

^cRace and ethnicity were assessed because of evidence from the available literature suggesting an association with outcomes in pediatric patients treated for appendicitis. Related data were reported by caregivers using surveys with predefined categories listed in the table. Other includes Black, Hispanic, Asian, American Indian, Alaska Native, and biracial patients. For patients managed nonoperatively with complete follow-up, this included 163 non-Hispanic White patients, 15 non-Hispanic Black patients, 19 Hispanic patients, 83 patients who reported another race or ethnicity, and 4 patients who did not report race or ethnicity. For patients in the surgery group with complete follow-up, this included 321 non-Hispanic White patients, 12 non-Hispanic Black patients, 46 Hispanic patients, 137 patients who reported another race or ethnicity, and 6 patients who did not report race or ethnicity.

Discussion

The success rate of nonoperative management and its association with fewer disability days than surgery for uncomplicated appendicitis in children did not differ by age, household income, or transfer status. However, the prespecified 70% success rate threshold for nonoperative management was not met for most subgroups.

Generalizability of these results may be limited, as only 19.3% of patients met eligibility criteria and unmeasured differences in patient characteristics may have introduced bias.

Section Editor: Jody W. Zylke, MD, Deputy Editor.

Supplement 1.

Protocol for the Initial Study

Supplement 2.

Nonauthor Collaborators. Midwest Pediatric Surgery Consortium