Abstract
Background
There are few objective data to guide management of cystic fibrosis (CF) pulmonary exacerbations. We studied intravenous (IV) antibiotic treatment failure as defined by a need to retreat patients with IV antibiotics within 30 days of completion of a prior IV antibiotic treatment for pulmonary exacerbation.
Methods
The first IV-treated exacerbation on or after Jan 1, 2010 among US CF Foundation Patient Registry patients was studied, combining treatments separated by <7 days into single treatments. IV treatment duration categories were: 1-4, 5-8, 9-12, 13-16, 17-22, and ≥23 days (inclusive). Logistic regressions for IV retreatment in ≤30 days were adjusted with 12 categorical covariates, including age, sex, lung function, prior-year exacerbations, CF complications, CF Care Program, and ever/never treated in hospital.
Results
777 of 13579 patients (5.7%) were retreated within 30 days, with incidence varying by treatment duration: 1-4 days, 8.7%; 5-8 days; 6.6%; 9-12 days, 3.2%; 13-16 days, 4.5%; 17-22 days, 6.2%; ≥23 days, 10.3% and hospitalization: ever, 5.0%; never 8.5%. Adjusted odds ratios (OR) for retreatment (compared to 13-16 days treatment) were: 1-4 days, 1.94 [95%CI 1.49, 2.54] P< .001; 5-8 days, 1.55 [1.18, 2.04] P=.002; 9-12 days, 0.78 [0.58, 1.04] P=.09; 17-22 days, 1.12 [0.88, 1.42] P=.37; ≥23 days, 1.46 [1.12, 1.91] P=.005. Adjusted retreatment OR for never/ever hospitalized was 1.57 [1.29, 1.90] P<.001. Prior-year exacerbation number, oxygen therapy, non-invasive ventilation, and female sex were significantly associated with retreatment. Modeling hazard rate time-dependence showed that treatment duration and location-associated hazard rates attenuated within a few months after treatment.
Conclusion
After adjustment for covariates known to be associated with increased risk of IV treatment for exacerbation, IV antibiotic treatments of <9 and ≥23 days and those without hospitalization were significant risk factors for IV retreatment within 30 days of completion of an exacerbation treatment.
Keywords: pulmonary exacerbation, treatment failure, risk factors
Graphical abstract
BACKGROUND
Pulmonary exacerbations are important clinical events in the lives of people with cystic fibrosis (CF) [1, 2], with those treated with intravenous (IV) antibiotics for exacerbation experiencing reduced quality of life [3], increased resource utilization [4, 5], accelerated lung disease progression [6, 7] and increased mortality risk [8-11]. With few objective data to guide practice, there is tremendous variability in exacerbation management within and across CF Care Programs [12-15]. Given exacerbation care diversity, there is a real possibility that some treatment approaches may be more effective than others, either by achieving better results or by achieving equivalent results with reduced resource utilization or toxicity risk [2]. However, attempts to compare the effectiveness of different treatment approaches using observational data [16-18] have been hampered by strong indication bias: sicker patients tend to be treated more aggressively than healthier patients, and in turn comparatively worse outcomes can be associated with more aggressive treatments [19].
In recent analyses of the association between IV antibiotic-treated exacerbations in a prior year and hazard of future exacerbation among patients followed in the US CF Foundation Patient Registry (CFFPR)[20, 21], we noted two treatment-associated covariates that appeared to have strong association with elevated future exacerbation hazard and that could not be readily explained by indication bias: IV treatments of less than 9 days in duration and treatments in which patients were never admitted to hospital. These observations suggest that shorter IV antibiotic treatments and those in which there is no hospital admission, both of which are care decisions made at the time of exacerbation, may be inferior management practices leading to poorer outcomes. In the current observational analysis, we further characterize the covariates of abbreviated IV antibiotic treatment and outpatient treatment without hospital admission with respect to the CF populations likely to have received this type of management, whether these care patterns are associated with poorer near-term health outcomes as assessed by the need for IV-antibiotic retreatment within 30 days, and the level of evidence that indication bias has affected these relationships.
METHODS
The CFFPR population data set used for these studies has been described previously [21]. Briefly, to be included, CFFPR patients had to have been treated with IV antibiotics for exacerbation at least once on or after January 1, 2010, and to have had at least one additional clinic visit or a death recorded by December 31, 2013. Patients also had to have data fields related to clinical complications completed in the CFFPR case report form during the calendar year of their exacerbation treatment. Exacerbations and clinic visits occurring after lung transplantation were excluded from analyses. Adults and guardians of children followed in the CFFPR provided consent for data analyses. CFFPR patient data were de-identified prior to analyses, which were approved prospectively by the CFFPR Registry Committee and the University Hospitals Case Medical Center Institutional Review Board (#01-15-15). Statistical analyses were performed with R and MedCalc Statistical Software (MedCalc Software bvba, Ostend, Belgium).
IV-antibiotic treatments identified as for pulmonary exacerbation that were separated by less than 7 days were concatenated and considered single treatments in order to avoid mislabeling small administrative gaps in single management episodes (e.g., transfer from hospital- to home-IV treatment over a weekend) as multiple independent exacerbation events as previously described [21]. IV antibiotic treatment durations were categorized in 4-day increments up to and beyond 14 days (i.e., 1-4 days, 5-8 days, 9-12 days, and 13-16 days); with two additional categories for ‘longer’ IV treatments (17-22 days) and ‘extended’ IV treatments (≥23 days). Deciles of IV-treatment duration were studied in sensitivity analyses. Admission to hospital was treated as a binary (yes/no) variable regardless of duration or proportion of hospital stay. Patient demographic covariates were collected from CFFPR Annual Review data and categorized as described previously [21].
Associations of IV antibiotic treatment duration and hospital admission with IV antibiotic retreatment for exacerbation ≥7 days and ≤30 days after the end of IV treatment were studied using logistic regression. Each covariate was studied independently and then in a combined model that also included additional covariates previously shown to be significantly associated with elevated exacerbation hazard within 6 months of the end of treatment [21], as well as a categorical variable for the 261 US CF Care Programs contributing data to the analysis. For treatment duration categories, odds of retreatment were compared to the category containing the most frequent treatment duration. Patients classified as not retreated and with a last post-exacerbation encounter date within 30 days of the end of exacerbation treatment (i.e. those lacking a complete 30-day follow-up period) were considered to have not been treated within 30 days for the purposes of logistic regression modeling. These same patients were counted as having been retreated within 30 days in subsequent sensitivity analyses.
Variation of covariate association with future exacerbation hazard with time from the end of exacerbation treatment was studied using a Cox proportional hazards model with time-varying coefficients. The model included covariates shown to be significantly associated with IV retreatment of exacerbations within 30 days of the end of exacerbation treatment, and was repeatedly fit allowing one covariate at a time to have a time-varying coefficient while adjusting for the remaining covariates. A 7 degree-of-freedom natural cubic spline with knots evenly distributed from 0 to 24 months was used to model time dependence. The proportional hazard assumption was made for events exceeding 24 months after end of treatment by fixing the first derivative of the spline to 0 for times greater than 24 months (reducing degrees of freedom by 1).
RESULTS
Among 13,579 CFFPR patients treated with IV antibiotics for pulmonary exacerbation on or after January 1, 2010, 1375 (10.1%) were treated for 1-4 days, 1416 (10.4%) were treated for 5-8 days, 2116 (15.6%) were treated for 9-12 days, 5395 (39.7%) were treated for 13-16 days, 2177 (16.0%) were treated for 17-22 days, and 1200 (8.8%) were treated for at least 23 days (Table 1). Most patients (10,924; 80.4%) had been admitted to hospital at some point during their IV antibiotic treatment. There were clear demographic trends across IV antibiotic treatment duration categories; younger patients and those with better lung function were more likely to be treated for shorter durations with IV antibiotics while older patients and those with more advanced lung disease were more likely to be treated for longer durations (Table 1). Similarly, patients who had received fewer IV treatments for pulmonary exacerbation in the prior year and those who did not require oxygen therapy or non-invasive ventilation were more likely to be treated for comparatively shorter durations with IV antibiotics. Although a majority of patients were admitted to hospital at some point during IV antibiotic treatment, relatively fewer older patients and patients with advanced lung disease were admitted during treatment compared with younger, healthier patients (Table 1). Little difference was observed in the proportion of patients hospitalized at any time during IV antibiotic treatment as a function of prior-year IV-treated exacerbations, and patients who required oxygen therapy or non-invasive ventilation were only marginally more likely to be admitted to hospital at some point during treatment (Table 1). Relationships between hospital admission and IV antibiotic treatment durations were more complicated, with relatively fewer patients at IV antibiotic treatment duration extremes (treated 1-4 days or treated 17 or more days) being admitted to hospital compared to patients treated for between 5 days and 16 days with IV antibiotics (Table 1, Figure 1A).
Table 1.
Patient Demographics by IV Antibiotic Treatment Duration and Hospital Admission
| Treated 1-4 Days (N=1375) | Treated 5-8 Days (N=1416) | Treated 9-12 Days (N=2116) | Treated 13-16 Days (N=5295) | Treated 17-22 Days (N=2177) | Treated ≥23 Days (N=1200) | Admitted to Hospital (N=10924) | Not Admitted to Hospital (N=2655) | |
|---|---|---|---|---|---|---|---|---|
| Age group, N (%)* | ||||||||
| <6 yrs | 289 (19.4%) | 228 (15.3%) | 292 (19.6%) | 529 (35.6%) | 105 (7.1%) | 45 (3.0%) | 1256 (84.4%) | 232 (15.6%) |
| 6-12 yrs | 257 (9.5%) | 272 (10.1%) | 502 (18.6%) | 1107 (40.9%) | 399 (14.8%) | 167 (6.2%) | 2346 (86.8%) | 358 (13.2%) |
| 13-17 yrs | 239 (9.2%) | 274 (10.5%) | 485 (18.6%) | 1018 (39.1%) | 400 (15.4%) | 189 (7.3%) | 2248 (86.3%) | 357 (13.7%) |
| 18-24 yrs | 262 (8.9%) | 336 (11.4%) | 442 (15.0%) | 1115 (37.8%) | 519 (17.6%) | 279 (9.4%) | 2367 (80.2%) | 586 (19.8%) |
| 25-50 yrs | 238 (8.3%) | 223 (7.8%) | 302 (10.5%) | 1153 (40.2%) | 567 (19.8%) | 385 (13.4%) | 2068 (72.1%) | 800 (27.9%) |
| >50 yrs | 90 (9.4%) | 83 (8.6%) | 93 (9.7%) | 373 (38.8%) | 187 (19.5%) | 135 (14.0%) | 639 (66.5%) | 322 (33.5%) |
| Sex, N (%) | ||||||||
| Male | 672 (10.3%) | 714 (10.9%) | 1028 (15.7%) | 2549 (39%) | 1022 (15.6%) | 547 (8.4%) | 5316 (81.4%) | 1216 (18.6%) |
| Female | 703 (10.0%) | 702 (10.0%) | 1088 (15.4%) | 2746 (39%) | 1155 (16.4%) | 653 (9.3%) | 5608 (79.6%) | 1439 (20.4%) |
| FEV1 % predicted group, N (%) | ||||||||
| NA, <6yrs | 267 (20.6%) | 197 (15.2%) | 256 (19.7%) | 459 (35.3%) | 85 (6.5%) | 35 (2.7%) | 1085 (83.5%) | 214 (16.5%) |
| ≥100 | 191 (13.1%) | 195 (13.4%) | 307 (21.1%) | 531 (36.5%) | 152 (10.5%) | 77 (5.3%) | 1225 (84.3%) | 228 (15.7%) |
| 70-<100% | 491 (9.8%) | 521 (10.4%) | 856 (17.0%) | 2063 (41.0%) | 755 (15.0%) | 341 (6.8%) | 4117 (81.9%) | 910 (18.1%) |
| 40-<70% | 312 (7.7%) | 336 (8.3%) | 540 (13.4%) | 1582 (39.2%) | 789 (19.5%) | 479 (11.9%) | 3147 (77.9%) | 891 (22.1%) |
| <40% | 86 (5.6%) | 133 (8.6%) | 126 (8.2%) | 587 (38.1%) | 365 (23.7%) | 245 (15.9%) | 1168 (75.7%) | 374 (24.3%) |
| NA, ≥6yrs | 28 (12.7%) | 34 (15.5%) | 31 (14.1%) | 73 (33.2%) | 31 (14.1%) | 23 (10.5%) | 182 (82.7%) | 38 (17.3%) |
| IV-Treated Exacerbations in Prior Year | ||||||||
| None | 935 (11.5%) | 959 (11.8%) | 1395 (17.2%) | 3155 (39.0%) | 1117 (13.8%) | 537 (6.6%) | 6611 (81.6%) | 1487 (18.4%) |
| One | 231 (9.4%) | 197 (8.0%) | 308 (12.5%) | 994 (40.5%) | 473 (19.3%) | 254 (10.3%) | 1889 (76.9%) | 568 (23.1%) |
| Two | 117 (8.0%) | 122 (8.3%) | 193 (13.2%) | 580 (39.6%) | 270 (18.4%) | 184 (12.6%) | 1165 (79.5%) | 301 (20.5%) |
| Three | 45 (5.9%) | 64 (8.4%) | 109 (14.3%) | 285 (37.4%) | 151 (19.8%) | 108 (14.2%) | 618 (81.1%) | 144 (18.9%) |
| ≥Four | 47 (5.9%) | 74 (9.3%) | 111 (13.9%) | 281 (35.3%) | 166 (20.9%) | 117 (14.7%) | 641 (80.5%) | 155 (19.5%) |
| Oxygen Therapy | ||||||||
| No | 1162 (10.6%) | 1156 (10.5%) | 1842 (16.7%) | 4336 (39.4%) | 1667 (15.2%) | 840 (7.6%) | 8820 (80.2%) | 2183 (19.8%) |
| Yes | 154 (6.8%) | 205 (9.0%) | 232 (10.2%) | 874 (38.5%) | 469 (20.6%) | 339 (14.9%) | 1891 (83.2%) | 382 (16.8%) |
| Non-Invasive Ventilation | ||||||||
| No | 1289 (10.0%) | 1323 (10.3%) | 2032 (15.8%) | 5061 (39.4%) | 2040 (15.9%) | 1089 (8.5%) | 10343 (80.6%) | 2491 (19.4%) |
| Yes | 26 (6.4%) | 36 (8.8%) | 38 (9.3%) | 134 (32.9%) | 88 (21.6%) | 85 (20.9%) | 336 (82.6%) | 71 (17.4%) |
| Admitted to Hospital | ||||||||
| Yes | 735 (6.7%) | 1212 (11.1%) | 1922 (17.6%) | 4428 (40.5%) | 1719 (15.7%) | 908 (8.3%) | ||
| No | 640 (24.1%) | 204 (7.7%) | 194 (7.3%) | 867 (32.7%) | 458 (17.3%) | 292 (11.0%) | ||
percentages for treatment duration and hospital admission are across rows
NA= not available
Figure 1. IV Antibiotic Treatment Durations, Hospital Admissions, and IV Retreatment within 30 Days.
Panel A) Distribution of IV antibiotic treatment durations. Black bars show numbers of patients of a given treatment duration who were never admitted to hospital as part of their exacerbation treatment. Panel B) Proportions of patients retreated with IV antibiotics within 30 days of the end of exacerbation treatment by original IV treatment duration category (gray) or hospital admission category (white). Bars are 95% confidence intervals.
Among the 96.1% of patients (N=13,058) who had a routine clinic visit reported after the end of their IV antibiotic treatment and before a second IV-treated pulmonary exacerbation, the median time from end of treatment to this visit was 34 days, with 45.4% (5,932 patients) having visited clinic by 30 days after the end of their IV-antibiotic treatment. In all, 5.7% of patients (777 of 13,579) were treated again with IV antibiotics for pulmonary exacerbation ≥7 days and ≤30 days after the end of their initial IV antibiotic treatments for exacerbation; 183 patients (1.0%) were not retreated but lacked an entire 30-day follow-up period after exacerbation treatment. Proportions of retreated patients varied substantially among IV treatment duration subgroups, ranging from 3.2% to 10.3% (Figure 1B). A lower proportion of patients who received at least some portion of their IV antibiotic treatment in hospital were retreated with IV antibiotics within 30 days when compared with those patients that were never admitted to hospital (5.0% versus 8.5%; Figure 1B).
Increased odds of retreatment remained significant for patients treated <9 days and ≥23 days when compared with those treated 13-16 days after adjustment for admission to hospital, CF Care Program, and 13 covariates previously shown to be significantly associated with future exacerbation hazard [21] (Figure 2, Table 2). Similarly, patients never admitted to hospital had higher odds of retreatment compared to patients ever admitted to hospital after adjustment (Figure 2, Table 2). In contrast, adjusted odds of retreatment in 30 days among patients treated 9-12 days and 17-22 days were not significantly different than those of patients treated 13-16 days. Only 6 of 16 covariates included in multivariate modeling (IV treatment duration, hospital admission, number of prior-year IV-treated exacerbations, oxygen therapy, non-invasive ventilation, and sex) were significantly associated with increased odds of IV antibiotic retreatment within 30 days (Table 2). These relationships were unaffected in sensitivity analyses in which the 138 patients without a complete 30-day follow-up were assumed to have been retreated for exacerbation within 30 days or by categorization of treatment duration in deciles.
Figure 2. Logistic Regression Modeling of IV Antibiotic Retreatment within 30 days using IV Antibiotic Treatment Duration and Hospital Admission Covariates.
Open circles show univariate odds ratios of IV antibiotic retreatment. Closed circles show odds ratios derived from multivariate modeling that included CF Care Program and 13 other covariates previously shown to be significantly associated with elevated hazard of future pulmonary exacerbation (Table 2). Unadjusted odds ratios are shown above point estimates and adjusted odds ratios are shown below estimates. Bars show 95% confidence intervals.
Table 2.
Multivariate Logistic Regression for IV-Antibiotic Treatment ≥7 Days and ≤30 Days after End of IV Antibiotic Treatment for Exacerbation
| Covariate* | Regression Coefficient ± SE | Retreatment Odds Ratio [95% CI] | Regression P-Value |
|---|---|---|---|
| IV Treatment Duration (Days), versus 13 to 16 Days | |||
| 1 to 4 | 0.67 ± 0.14 | 1.94 [1.49, 2.54] | <0.001 |
| 5 to 8 | 0.44 ± 0.14 | 1.55 [1.18, 2.04] | 0.002 |
| 9 to 12 | −0.25 ± 0.15 | 0.78 [0.58, 1.04] | 0.093 |
| 17 to 22 | 0.11 ± 0.12 | 1.12 [0.88, 1.42] | 0.37 |
| ≥23 | 0.38 ± 0.14 | 1.46 [1.12, 1.91] | 0.005 |
| Hospital Admission, versus Ever | |||
| Never | 0.45 ± 0.1 | 1.57 [1.29, 1.90] | <0.001 |
| Prior-Year IV-Treated Exacerbations, versus No Prior-Year Treatments | |||
| One | 0.13 ± 0.12 | 1.13 [0.9, 1.43] | 0.29 |
| Two | 0.46 ± 0.13 | 1.58 [1.23, 2.03] | <0.001 |
| Three | 0.76 ± 0.15 | 2.14 [1.60, 2.88] | <0.001 |
| Four or more | 1.7 ± 0.13 | 5.47 [4.28, 7.00] | <0.001 |
| FEV1 % Predicted Group, versus ≥100% predicted | |||
| 70 to <100 | −0.27 ± 0.16 | 0.76 [0.56, 1.05] | 0.092 |
| 40 to <70 | −0.07 ± 0.17 | 0.93 [0.67, 1.29] | 0.66 |
| <40 | −0.20 ± 0.20 | 0.82 [0.56, 1.20] | 0.30 |
| Unknown, age<6yrs | 0.13 ± 0.20 | 1.14 [0.77, 1.67] | 0.51 |
| Unknown, age ≥6yrs | −0.01 ± 0.31 | 0.99 [0.53, 1.83] | 0.97 |
| Sex, versus Male | |||
| Female | 0.23 ± 0.08 | 1.26 [1.07, 1.48] | 0.005 |
| Smoking Status, versus Non-Smoker | |||
| Smoker | 0.25 ± 0.25 | 1.29 [0.79, 2.09] | 0.31 |
| Unknown | 0.14 ± 0.18 | 1.15 [0.81, 1.63] | 0.44 |
| Oxygen Therapy, versus No | |||
| Yes | 0.88 ± 0.11 | 2.42 [1.96, 2.98] | <0.001 |
| Unknown | 1.09 ± 0.36 | 2.96 [1.46, 6.00] | 0.003 |
| Non-Invasive Ventilation, versus No | |||
| Yes | 0.66 ± 0.16 | 1.93 [1.40, 2.66] | <0.001 |
| Unknown | −0.59 ± 0.38 | 0.55 [0.26, 1.17] | 0.12 |
| Allergic Bronchopulmonary Aspergillosis, versus No | |||
| Yes | 0.13 ± 0.14 | 1.14 [0.86, 1.51] | 0.38 |
| Sinus Disease, versus No | |||
| Yes | 0.16 ± 0.10 | 1.17 [0.97, 1.42] | 0.10 |
| Gastroesophageal Reflux Disease, versus No | |||
| Yes | 0.11 ± 0.10 | 1.11 [0.92, 1.34] | 0.26 |
| Glucose Tolerance, versus Normal | |||
| Impaired Glucose Tolerance | −0.11 ± 0.15 | 0.90 [0.67, 1.19] | 0.45 |
| CF-Related Diabetes | 0.12 ± 0.10 | 1.13 [0.93, 1.37] | 0.21 |
| Unknown | −0.19 ± 0.53 | 0.83 [0.29, 2.36] | 0.72 |
| Liver Cirrhosis, versus No | |||
| Yes | 0.33 ± 0.2 | 1.40 [0.94, 2.07] | 0.097 |
| Pancreatitis, versus No | |||
| Yes | 0.41 ± 0.24 | 1.50 [0.95, 2.38] | 0.084 |
| P. aeruginosa Isolation versus No | |||
| Yes | 0.11 ± 0.09 | 1.11 [0.93, 1.33] | 0.26 |
| Unknown | −0.06 ± 0.41 | 0.94 [0.42, 2.12] | 0.88 |
| Constant | −3.46 | <0.001 | |
Regression coefficients and retreatment odds ratios not shown for the 261 CF Care Programs included as a categorical variable in the model.
Among the six statistically significant covariates, hazard ratios for four (number of prior-year IV-treated exacerbations, oxygen therapy, non-invasive ventilation, and sex) did not substantially decrease over the 6 months after end of IV treatment (Figure 3). In contrast, hazards of exacerbation associated with the two treatment-related covariates (IV treatment duration and treatment in hospital) appeared to steadily decrease in importance after treatment cessation and were negligible within two to three months after treatment cessation (Figure 3).
Figure 3. Time-variation of covariate association with future exacerbation hazard from end of exacerbation treatment.
Hazard ratios for future exacerbation (solid lines) and associated 95% confidence intervals (shaded areas) over the 6 months immediately following IV antibiotic treatment cessation. Values are adjusted by inclusion of all covariates shown to be significantly associated with retreatment within 30 days of end of exacerbation treatment. PEx = pulmonary exacerbation
DISCUSSION
Health outcomes associated with CF pulmonary exacerbations are sufficiently poor and approaches to exacerbation management are sufficiently diverse that efforts to comb through observational data for evidence of more or less effective management practices seem justified. Yet there are fundamental problems with such analyses: despite wild diversity of care, patients are not randomly allocated to receive different treatments by their healthcare providers, and for this reason we cannot assume that observed differences in outcome associated with different treatments are caused by treatment differences. Treatment variability appears embedded in CF exacerbation treatment guidelines, which note a minimum 10-day treatment will “often be extended” based on empiric observations including extent of clinical improvement, changes in pulmonary function tests, patient history, and disease severity, and that treatment should proceed for 3 to 4 days beyond return to “baseline pulmonary signs and symptoms” [22]. A more recent objective literature review by the US CF Foundation concluded that “there is insufficient evidence to recommend an optimal duration of antibiotic treatment” for CF exacerbation [23].
We have observed that two exacerbation management treatment decisions, duration of IV antibiotic treatment and treatment location (never versus ever admitted to hospital), are strongly associated with retreatment with IV antibiotics within 30 days of the end of initial treatment. It is not possible to discriminate between IV retreatment due to relapse secondary to inadequate management (i.e. ‘treatment failure’) versus occurrence of a ‘new’ exacerbation in a given individual. However, it is reasonable to assume that treatment for a new event within 30 days is more likely to occur in individuals with higher underlying risks for exacerbation, such as those patients with more numerous IV-treated exacerbations in the prior year, decreased lung function and/or increased pulmonary complications [21]. By extension, we would predict that treatment for new events within 30 days is less likely to occur in other, ‘healthier’ patients.
In our study, extended IV antibiotic treatments were commonly chosen for older patients, patients with more advanced lung disease, and those with a history of prior-year IV-treated exacerbations, all groups at increased risk for future exacerbation. It is not surprising that we saw an increased risk of retreatment within 30 days for these patients compared with patients treated for 13 to 16 days, as noted above. In contrast, IV antibiotic treatments of <9 days (chosen for ~20% of CFFPR patients) were more commonly administered to patients with substantially lower risk of IV treatment for exacerbation: children under 6 years of age, individuals with better lung function, and those with few or no exacerbations treated with IV antibiotics in the prior year [21]. Given that these patients were at an elevated risk of retreatment in 30 days but were at a relatively low underlying exacerbation risk, it follows that a majority of retreatment events in this population must have been related to inadequate management of the initial exacerbation.
Whereas hazard ratios associated with significant demographic and complication risk covariates (prior-year exacerbation number, oxygen therapy, noninvasive ventilation, and sex) remained relatively constant over the 6 months following the end of exacerbation treatment, the same could not be said for treatment-related covariates (IV treatment duration and treatment location), which became attenuated within a few months of treatment cessation (Figure 3). Thus, how long and where a patient was treated strongly influenced his or her hazard of ‘near-term’ retreatment (i.e., treatment failure), but apparently had less influence on longer-term exacerbation hazard, an observation that is easier to reconcile with a more direct association between treatment variability and retreatment risk than to underlying indication bias.
Among the 261 Care Programs with treatment events in our analysis, an average of 20.7% of events per Program (SD=17.1%; median = 15.0%) did not include hospital admission. All but 19 Care Programs (7.3%) had at least one treatment event that did not include hospital admission and only 7 Programs (2.7%) had more than 60% of treatment events that did not include hospital admission. Proportion of treatments including hospital admission was not influenced by the number of events observed within a Program. Relationships between hospital admission and retreatment in 30 days were more complicated to interpret than treatment duration, in part because of the unusual distribution of the covariate in the population. An assumption that admission to hospital might be more common for patients with advanced disease and thus be associated with poorer outcomes due to indication bias is not supported by demographic data. In fact, older patients and patients with more advanced lung disease (patients more likely to receive extended IV antibiotic treatments) were less likely to be admitted to hospital in association with IV antibiotic treatment. However, proportionally more patients treated for short periods were not admitted to hospital than those treated for longer periods, complicating this relationship. Although it is difficult with these data to conclude that hospital admission itself reduces risk of retreatment in 30 days as opposed to being confounded by indication bias, it is important to note again that exacerbation hazard associated with treatment outside of hospital appears to attenuate with time elapsed from treatment end, an observation that would be consistent with a more direct effect of outpatient-only treatment on retreatment odds.
There are limitations to observational analyses employing disease registries, including an assumption of accurate and complete data capture by contributors. For this analysis, we prospectively established IV antibiotic administrations separated by less than 7 days as components of the ‘same’ treatment to avoid mistaking brief treatment gaps within a single management episode for multiple independent events. Although we feel this was a conservative approach for characterizing treatment, it had the potential effect of reducing the number of very short antibiotic treatments captured, and also may have missed ‘treatment failures’ occurring before 7 days after end of IV treatment. A small fraction of patients classified as not having been retreated within 30 days lacked complete 30-day follow-up data, and it was not possible to determine their 30-day outcomes with certainty. However, it seems unlikely that many of these patients were immediately retreated and simultaneously lost to follow-up, and sensitivity analyses in which all such patients were assumed to have been retreated within 30 days did not affect our results. There are alternative approaches for capturing a hospital admission covariate for logistic regression analyses, including counting total days of treatment in hospital or proportions of treatments received in hospital; each can be confounded by treatment duration. For instance, patients treated for shorter periods were likely to have fewer days of treatment in hospital, but may have greater relative proportions of treatment in hospital, when compared with patients treated for longer durations. An important covariate that was unavailable for our analyses was antibiotic treatment choice. Presumably, there are ‘better’ and ‘worse’ choices that can be made with respect to which antibiotics to prescribe for a given patient, as well as the regimen and dose of a given antibiotic administered, and these choices also may contribute to risk of retreatment within 30 days.
Despite these limitations, our results suggest that shorter IV antibiotic treatments for exacerbation place CF patients at greater risk for retreatment with IV antibiotics within 30 days, and that this elevated risk appears to relate to inadequate treatment of the initial exacerbation. The justification(s) for administration of shorter IV treatments remain unclear and are likely multifaceted. It may be that some patients experienced very rapid lung function recovery or symptom reduction that met clinician goals, that risk/benefit assessments favored shorter treatments in these healthier patients, that economic or psychosocial considerations forced the hand of treating clinicians, or a that combination of these and/or other factors conspired. Interestingly, IV treatment durations among patients with Medicaid insurance were comparable to patients with other insurance coverage, those with Medicaid insurance were significantly more likely than those with known non-Medicaid insurance to be admitted to hospital for some portion of their treatment (85.4% vs 77.4%, P<.0001), and Medicaid insurance coverage was not associated with increased odds of retreatment within 30 days. Given that the 30-day retreatment risk for patients receiving abbreviated exacerbation treatment regimens was almost twice that of patients receiving more substantial treatments, justifications for abbreviated regimens should be carefully examined.
ACKNOWLEDGEMEMTS
The authors thank the patients, families, CF care providers, and support staff associated with the Cystic Fibrosis Foundation Patient Registry, as well as Dr. Nicole Mayer-Hamblett for helpful discussions of analyses of hazard ratio time dependence. This project was supported by the Cystic Fibrosis Foundation (KONSTA09Y0), the National Institutes of Health through the Clinical and Translational Science Collaborative of Cleveland (UL1TR000439) and P30 DK027651, and the South Carolina Clinical & Translational Research (SCTR) Institute, with an academic home at the Medical University of South Carolina through National Institutes of Health grant UL1TR001450.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Ferkol T, Rosenfeld M, Milla CE. Cystic fibrosis pulmonary exacerbations. J Pediatr. 2006 Feb;148(2):259–64. doi: 10.1016/j.jpeds.2005.10.019. Review. [DOI] [PubMed] [Google Scholar]
- 2.Flume PA, VanDevanter DR. Pulmonary exacerbations. In: Bush A, Bilton D, Hodson M, editors. Hodson and Geddes' Cystic Fibrosis. 4th Edition. CRC Press; 2015. [Google Scholar]
- 3.Britto M, Kotagal U, Hornung R, Atherton H, Tsevat J, Wilmott R. Impact of recent pulmonary exacerbations on quality of life in patients with cystic fibrosis. Chest. 2002;121:64–72. doi: 10.1378/chest.121.1.64. [DOI] [PubMed] [Google Scholar]
- 4.Lieu T, Ray G, Farmer G, Shay G. The cost of medical care for patients with cystic fibrosis in a health maintenance organization. Pediatrics. 1999;103:e72. doi: 10.1542/peds.103.6.e72. [DOI] [PubMed] [Google Scholar]
- 5.Ouyang L, Grosse SD, Amendah DD, Schechter MS. Healthcare expenditures for privately insured people with cystic fibrosis. Pediatr Pulmonol. 2009;44(10):989–96. doi: 10.1002/ppul.21090. [DOI] [PubMed] [Google Scholar]
- 6.Konstan M, Morgan W, Butler S, Pasta D, Craib M, Silva S, Stokes D, Wohl M, Wagener J, Regelmann W, et al. Risk factors for rate of decline in forced expiratory volume in one second in children and adolescents with cystic fibrosis. J Pediatr. 2007;151:134–139. 139. doi: 10.1016/j.jpeds.2007.03.006. [DOI] [PubMed] [Google Scholar]
- 7.VanDevanter DR, Wagener JS, Pasta DJ, Elkin E, Jacobs JR, Morgan WJ, Konstan MW. Pulmonary outcome prediction (POP) tools for cystic fibrosis patients. Pediatr Pulmonol. 2010;45:1156–66. doi: 10.1002/ppul.21311. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Liou T, Adler F, Fitzsimmons S, Cahill B, Hibbs J, Marshall B. Predictive 5-year survivorship model of cystic fibrosis. Am J Epidemiol. 2001;153:345–352. doi: 10.1093/aje/153.4.345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Mayer-Hamblett N, Rosenfeld M, Emerson J, Goss C, Aitken M. Developing cystic fibrosis lung transplant referral criteria using predictors of 2-year mortality. Am J Respir Crit Care Med. 2002;166:1550–1555. doi: 10.1164/rccm.200202-087OC. [DOI] [PubMed] [Google Scholar]
- 10.Emerson J, Rosenfeld M, McNamara S, Ramsey B, Gibson RL. Pseudomonas aeruginosa and other predictors of mortality and morbidity in young children with cystic fibrosis. Pediatr Pulmonol. 2002;34(2):91–100. doi: 10.1002/ppul.10127. [DOI] [PubMed] [Google Scholar]
- 11.Ellaffi M, Vinsonneau C, Coste J, Hubert D, Burgel PR, Dhainaut JF, et al. One–year outcome after severe pulmonary exacerbation in adults with cystic fibrosis. Am J Respir Crit Care Med. 2005;171(2):158–64. doi: 10.1164/rccm.200405-667OC. [DOI] [PubMed] [Google Scholar]
- 12.Johnson C, Butler SM, Konstan MW, Morgan W, Wohl ME. Factors influencing outcomes in cystic fibrosis: a center–based analysis. Chest. 2003;123(1):20–7. doi: 10.1378/chest.123.1.20. [DOI] [PubMed] [Google Scholar]
- 13.Kraynack NC, Gothard MD, Falletta LM, McBride JT. Approach to treating cystic fibrosis pulmonary exacerbations varies widely across US CF care centers. Pediatr Pulmonol. 2011;46(9):870–81. doi: 10.1002/ppul.21442. [DOI] [PubMed] [Google Scholar]
- 14.Heltshe SL, Goss CH, Thompson V, Sagel SD, Sanders DB, Marshall BC, Flume PA. Short–term and long–term response to pulmonary exacerbation treatment in cystic fibrosis. Thorax. 2016 Mar;71(3):223–9. doi: 10.1136/thoraxjnl-2014-206750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cystic Fibrosis Foundation Patient Registry . 2014 Annual data report to the Center Directors. Cystic Fibrosis Foundation; Bethesda Maryland: 2015. [Google Scholar]
- 16.Thornton J, Elliott R, Tully MP, Dodd M, Webb AK. Long term clinical outcome of home and hospital intravenous antibiotic treatment in adults with cystic fibrosis. Thorax. 2004 Mar;59(3):242–6. doi: 10.1136/thx.2003.005876. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nazer D, Abdulhamid I, Thomas R, Pendleton S. Home versus hospital intravenous antibiotic therapy for acute pulmonary exacerbations in children with cystic fibrosis. Pediatr Pulmonol. 2006 Aug;41(8):744–9. doi: 10.1002/ppul.20433. [DOI] [PubMed] [Google Scholar]
- 18.Collaco JM, Green DM, Cutting GR, Naughton KM, Mogayzel PJ., Jr Location and duration of treatment of cystic fibrosis respiratory exacerbations do not affect outcomes. Am J Respir Crit Care Med. 2010 Nov 1;182(9):1137–43. doi: 10.1164/rccm.201001-0057OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Schechter MS, Wagener J, Konstan MW, Morgan W. Location and duration of treatment of cystic fibrosis respiratory exacerbations. Am J Respir Crit Care Med. 2011 Jul 15;184(2):278. doi: 10.1164/ajrccm.184.2.278. [DOI] [PubMed] [Google Scholar]
- 20.VanDevanter DR, Pasta DJ, Konstan MW. Treatment and demographic factors affecting time to next pulmonary exacerbation in cystic fibrosis. J Cyst Fibros. 2015 Nov;14(6):763–769. doi: 10.1016/j.jcf.2015.02.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.VanDevanter DR, Morris NJ, Konstan MW. IV-treated pulmonary exacerbations in the prior year: An important independent risk factor for future pulmonary exacerbation in cystic fibrosis. J Cyst Fibros. 2015 Oct 22;:S1569–1993(15)00245-3. doi: 10.1016/j.jcf.2015.10.006. doi: 10.1016/j.jcf.2015.10.006. [Epub ahead of print] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Clinical Practice Guidelines for Cystic Fibrosis Committee . Clinical practice guidelines for cystic fibrosis. Cystic Fibrosis Foundation; Bethesda, MD: 1997. 1997. [Google Scholar]
- 23.Flume PA, Mogayzel PJ, Jr, Robinson KA, Goss CH, Rosenblatt RL, Kuhn RJ, Marshall BC. Clinical Practice Guidelines for Pulmonary Therapies Committee. Cystic fibrosis pulmonary guidelines: treatment of pulmonary exacerbations. Am J Respir Crit Care Med. 2009 Nov 1;180(9):802–8. doi: 10.1164/rccm.200812-1845PP. [DOI] [PubMed] [Google Scholar]