Key Findings and Strength of Evidence

This report reviews 59 randomized, controlled trials (RCTs) of treatments for seasonal allergic rhinitis (SAR) in adults and adolescents, in children younger than 12 years of age, and in pregnant women. In adults and adolescents, oral drug classes studied were selective and nonselective antihistamine, sympathomimetic decongestant, and leukotriene receptor antagonist; nasal drug classes were antihistamine, corticosteroid, and cromolyn. No RCTs or observational studies of intranasal anticholinergic or nasal saline spray were identified. In children, drug classes studied were oral selective and nonselective antihistamine. In this population, no RCTs or observational studies of nasal antihistamine, corticosteroid, or cromolyn; oral leukotriene receptor antagonist; or nasal saline spray were identified. No RCTs or observational studies of SAR treatments in pregnant women were identified.

Key Question 1. Comparative Effectiveness of SAR Treatments in Adults and Adolescents 12 Years of Age or Older

Overview of Results

Twenty-two treatment comparisons of interest were identified. We found studies that satisfied our inclusion criteria for 13 of these. Results for these comparisons are presented in Table 73 and discussed below. For most outcomes, evidence was insufficient to form any comparative effectiveness conclusion. In five comparisons, we found evidence for comparable effectiveness (equivalence) of treatments for at least one outcome (rows 5, 6, 8, 11, and 12 in Table 73), and we found evidence for superior effectiveness of one treatment over another for one outcome in each of two comparisons (row 5 and row 9 in Table 73).

Table 73

Summary of findings and strength of evidence for effectiveness in 13 treatment comparisons: Key Question 1–adults and adolescents.

When reviewing Table 73, it is important to keep in mind that the strength of evidence analysis only describes the evidence for each specific treatment comparison. That is, conclusions about equivalence or superiority can be made when two treatments are directly compared. In the absence of direct comparison, neither conclusion is supported. For example, for various nasal symptom outcomes, there was moderate strength evidence for comparable effectiveness (equivalence) of oral selective antihistamine and oral leukotriene receptor antagonist (row 5), and high strength evidence for the comparable effectiveness of intranasal corticosteroid and oral leukotriene receptor antagonist (row 8). This does not support a conclusion of equivalence of oral selective antihistamine and intranasal corticosteroid for nasal symptoms. As shown in row 3, direct evidence from the comparison of oral selective antihistamine to intranasal corticosteroid for the treatment of nasal symptoms was insufficient to form a conclusion about their comparative effectiveness. In contrast, high strength evidence suggests comparable effectiveness of intranasal corticosteroid plus nasal antihistamine combination therapy and each of its components for nasal and eye symptoms (rows 11 and 12). Direct evidence also suggests comparable effectiveness of intranasal corticosteroid and nasal antihistamine for these outcomes (row 6), suggesting comparable effectiveness of all three treatments.

It also is important to keep in mind that:

• Results presented in the summary table, indeed in the entire report, reflect the reporting of data in the literature. Data that were reported with insufficient detail to permit their inclusion in meta-analysis restricted the comparative effectiveness conclusions that could be drawn. Data reported with insufficient detail to assess their quality reduces the strength of the body of evidence. This was particularly evident in three-arm trials in which p-values for comparisons of interest for this review were not the primary comparisons in the trial. It may be that in some cases the reporting of the evidence, rather than the evidence itself, is insufficient to make any conclusion about the comparative effectiveness of two treatments. (See Limitations of the Evidence Base, below.)
• Seven of 13 treatment comparisons (54 percent) had poor representation (less than 50 percent) of at least one drug class compared. Although the Technical Expert Panel (TEP) was unaware of evidence suggesting differential effectiveness within a class, these seven comparisons may not adequately represent the classes of drugs compared, and conclusions are limited to the specific drugs studied. The seven comparisons are listed below with the proportion of drugs represented in each class indicated in parentheses.
  • Oral selective versus oral nonselective antihistamine (40 percent versus 25 percent)
  • Intranasal corticosteroid versus nasal antihistamine (25 percent versus 100 percent)
  • Intranasal corticosteroid versus oral leukotriene receptor antagonist (25 percent versus 100 percent)
  • Combination oral selective antihistamine plus intranasal corticosteroid versus oral selective antihistamine (40 percent of oral selective antihistamines and 25 percent of intranasal corticosteroids)
  • Combination oral selective antihistamine plus intranasal corticosteroid versus intranasal corticosteroid (60 percent of oral selective antihistamines and 25 percent of intranasal corticosteroids)
  • Combination intranasal corticosteroid plus nasal antihistamine versus intranasal corticosteroid (12.5 percent of intranasal corticosteroids and 50 percent of nasal antihistamines)
  • Combination intranasal corticosteroid plus nasal antihistamine versus nasal antihistamine (12.5 percent of intranasal corticosteroids and 50 percent of nasal antihistamines)

For some drug classes, the impact of poor representation may be limited; for example, 25 percent of oral nonselective antihistamines were studied in two comparisons, but this class of drugs is used less often since the advent of newer treatments. Similarly, only one of two oral decongestants was studied (pseudoephedrine), the one that is most commonly used.

In four comparisons, intranasal corticosteroid alone or in combination was poorly represented: In two (versus nasal antihistamine and versus oral leukotriene receptor antagonist), only fluticasone propionate and budesonide were studied. In the other two (combination intranasal corticosteroid plus nasal antihistamine versus each component), only fluticasone propionate was studied. Comparative effectiveness conclusions therefore apply to the specific drugs in each comparison; how well they generalize to other drugs in the same class is uncertain.

Cells marked “Insufficient” indicate insufficient evidence to form a conclusion. As described in the Methods section, conclusions that could be drawn for any outcome depended on the nature of the evidence available:

• For outcomes that had minimal clinically important differences (MCIDs) and meta-analysis was done, conclusions of superiority, equivalence, or insufficient evidence could be made.
• For outcomes that had an MCID and meta-analysis was not done, only conclusions of superiority or insufficient evidence could be made.
• For outcomes with no MCID, only conclusions of superiority or insufficient evidence could be made regardless of whether meta-analysis was done.

Of 28 meta-analyses conducted, 26 supported equivalence conclusions. The other two meta-analyses assessed oral selective antihistamine in comparison to intranasal corticosteroid (row 3). In approximately half of the trials identified for this comparison, treatment effects for nasal and eye symptoms were not reported. Therefore, the evidence was imprecise and insufficient to support a comparative effectiveness conclusion of superiority or equivalence for these outcomes. All other “Insufficient” cells indicate insufficient evidence to support only superiority conclusions.

As shown in Table 73, we found:

• High strength evidence for comparable effectiveness (equivalence) of:
  • Combination intranasal corticosteroid plus nasal antihistamine, intranasal corticosteroid monotherapy, and nasal antihistamine monotherapy for nasal and eye symptoms at 2 weeks.
  • Intranasal corticosteroid and oral leukotriene receptor antagonist (montelukast) for nasal symptoms at 2 weeks.
• Moderate strength evidence for comparable effectiveness of oral selective antihistamine and oral leukotriene receptor antagonist for nasal and eye symptoms and for improved quality of life at 2-4 weeks.
• Moderate strength evidence for the use of oral leukotriene receptor antagonist over oral selective antihistamine for reduced asthma rescue medication use at 2-4 weeks.
• Low strength evidence for the use of combination oral selective antihistamine plus intranasal corticosteroid over oral selective antihistamine monotherapy for improved quality of life at 2-4 weeks.

Key Question 2. Comparative Harms of SAR Treatments in Adults and Adolescents 12 Years of Age or Older

We identified two comparisons with sufficient evidence to support the use of one treatment over the other in order to avoid harm while treating SAR symptoms. These are shown in Table 74. To avoid insomnia at approximately 2 weeks, moderate strength evidence supported the use of oral selective antihistamine rather than either monotherapy with an oral decongestant or combination therapy with oral selective antihistamine plus oral decongestant. For all other adverse events of interest, evidence to indicate superior harms avoidance with one treatment compared to another was either insufficient or lacking. Because MCIDs for harms outcomes have not been defined, equivalence of treatments compared was not tested and cannot be assumed.

Table 74

Summary of findings and strength of evidence of harms in 13 treatment comparisons: Key Question 2–adults and adolescents.

As shown in Table 74, we made 46 harms assessments. Of these, 34 (74 percent) were based on drug class comparisons with less than 50 percent representation for at least one treatment compared. For these 34 assessments, conclusions may be limited to the specific drugs studied. As with effectiveness comparisons, the impact of poor representation may be limited for some drug classes (see “Overview of Results” for Key Question [KQ] 1, above).

We sought comparative information on a wide range of adverse events commonly associated with the pharmacologic classes studied. Conclusions about comparative harms are limited by the nature of the evidence reviewed in this report.

• Included trials lasted at most the duration of the pollen season, which is generally 6 to 8 weeks. Treatment durations were therefore relatively short for consideration of harms. Latent effects or cumulative effects from longer treatment durations are unknown.

Adverse events reported as the proportion of patients experiencing an event are assumed to be constant over time. This method of reporting does not differentiate between a person with a single episode of insomnia and one who experienced it every night of a 2-week trial. Adverse event frequencies often were not reported for all treatment groups. Harms assessment was inconsistent in the trials reviewed. Twenty-seven percent of trials indicated that an active method of harms surveillance was used. In contrast to passive surveillance, active surveillance of harms can yield qualitatively and quantitatively different results.¹⁴⁴ However, we had to assume consistency of harms surveillance across trials to synthesize estimates and to consider a body of evidence for comparative review. To mitigate the effects of inconsistent harms surveillance on pooled effects (i.e., meta-analysis), we considered for pooling only events that either were considered severe by investigators or resulted in treatment discontinuation. We found no candidates for meta-analysis in any comparison. Adverse events collected for this review were categorized as mild, moderate, or severe as they were identified in the source publication, with the exception that all adverse events leading to treatment discontinuation were considered severe.

Combined Evaluation of Key Questions 1 and 2. Comparative Effectiveness and Harms of Treatments in Adults and Adolescents 12 Years of Age or Older

We did not find evidence that any single treatment demonstrated both greater effectiveness and lower risk of harms. Table 75 shows the two comparisons for which there was sufficient evidence on reducing harms to form a conclusion along with the comparative effectiveness results for these comparisons. Moderate strength evidence supported the use of oral selective antihistamine to avoid insomnia associated with sympathomimetic decongestant at approximately 2 weeks (row 1 and row 2), but evidence was insufficient to support the use of one treatment over the other for effectiveness. (Equivalence was not assessed in either comparison due to the inability to conduct meta-analysis.) Similarly, of two treatments shown to be comparatively superior for effectiveness (row 3 and row 4), neither was preferred for harms avoidance.

Table 75

Comparison of efficacy and harms findings for two treatment comparisons.

Key Question 3. Comparative Effectiveness and Harms of SAR Treatments in Pregnant Women

For this KQ, we considered only Pregnancy Category B drugs, in which teratogenic effects have not been identified in animal studies or replicated in human studies. Evidence for the assessment of this KQ was lacking. No RCTs, observational studies, systematic reviews, or meta-analyses met the inclusion criteria.

Drugs used for the treatment of SAR have wide therapeutic windows (i.e., across the range of doses at which efficacy is seen, severe adverse events are not expected). Therefore, the choice of SAR treatment in pregnant women may be cautiously informed by comparative effectiveness evidence from the nonpregnant patient population. Because physiologic changes of pregnancy alter drug disposition, generalization of findings from the nonpregnant population to pregnant women requires knowledge of the magnitude and direction of these changes (Table 76). Due to a lack of study in pregnant women, current knowledge does not present a clear picture of safe and efficacious dosing adjustments of SAR treatments necessary to account for the physiologic changes of pregnancy.¹⁴⁵ The minimum effective dose is generally preferred pregnancy.

Table 76

Physiologic changes in pregnancy and potential effects on drug disposition.

Key Question 4. Comparative Effectiveness and Harms of SAR Treatments in Children Younger Than 12 Years of Age

Of 21 treatment comparisons of interest among children, studies that met our inclusion criteria were identified for one, selective versus nonselective oral antihistamine. No observational studies, systematic reviews, or meta-analyses met the required inclusion criteria.

The evidence for effectiveness and for harms was insufficient to support the use of either oral selective antihistamine or nonselective antihistamine for the treatment of nasal or eye symptoms in children younger than 12 years of age (mean age 9 years, range 4 to 12 years). This finding was based on trials that studied 20 percent of oral selective antihistamines and 9 percent of oral nonselective antihistamines used to treat children. As with harms outcomes, a finding of insufficient evidence to support a conclusion of superiority of one treatment over the other does not imply equivalence of the treatments. The evidence for benefit was truly insufficient; equivalence was not assessed.

Findings in Relationship to What Is Already Known

The following three systematic reviews provide current information about the pharmacologic treatment of allergic rhinitis. Each provided a description of the literature search, inclusion and exclusion criteria for identified trials, and quality assessments of included trials. Thus, the risk of bias was considered low for each. Two of the reviews were published before 2010, the cutoff date for potential incorporation of results into this review. Results from the third review were not incorporated into this review due to methodological differences; for example, the 2010 review was based on previously published systematic reviews, and disease was not limited to SAR. For purposes of comparing our findings to current knowledge, we included high-quality seminal works by relevant groups regardless of publication date.

• Guidelines for the treatment of allergic rhinitis from the international Allergic Rhinitis and Its Impact on Asthma (ARIA) working group, updated in 2010²⁸

A 2009 systematic review of treatments for hay fever¹⁴⁷

A 2008 Practice Parameter on the diagnosis and management of rhinitis from the Joint Task Force on Practice Parameters, representing the American Academy of Allergy, Asthma & Immunology (AAAAI), the American College of Allergy, Asthma and Immunology (ACAAI), and the Joint Council of Allergy, Asthma and Immunology (JCAAI)³

A 2010 systematic review of SAR treatments by drug class⁴¹ is not included in this list because quality assessments of included trials were not reported.

Findings from each of these reports are compared with those of this comparative effectiveness review in Table 77. Of 13 comparisons for which we found studies, three were not addressed by the systematic reviews. In two of the remaining ten comparisons, our conclusions agreed with at least one of the systematic reviews (ARIA guidelines in both instances). For five of eight discordant conclusions, other systematic reviews formed comparative effectiveness or harms conclusions, and we found insufficient evidence to do so. The other three discordant conclusions involved intranasal corticosteroid alone or in combination. We concluded comparable effectiveness (equivalence) of the treatments compared, and other systematic reviews concluded comparative superiority of intranasal corticosteroid. The eight discordant conclusions are reviewed below. In all cases, differing conclusions could be attributed to differences in inclusion criteria for trials reviewed.

Table 77

Comparison of findings from four systematic reviews of treatments for seasonal allergic rhinitis.

Discordant conclusions of insufficient evidence

• Oral selective antihistamine versus oral non-selective antihistamine: ARIA guidelines²⁸ and the AAAAI practice parameter³ recommended oral selective antihistamine to avoid harms. ARIA did not provide an evidence profile for this comparison. Evidence to inform AAAAI's recommendation included observational data and comparative studies in healthy volunteers.
• Oral selective antihistamine versus nasal antihistamine: ARIA guidelines²⁸ recommended oral selective antihistamine for avoidance of harms, such as bitter aftertaste. This review included trials of non-Food and Drug Administration (FDA) approved oral antihistamines (ebastine and levocabastine) and a trial in perennial allergic rhinitis (PAR). The other two reviews³, ¹⁴⁹ reported comparable efficacy of the treatments. One¹⁴⁷ of these conclusions was based on a trial of a non-FDA approved nasal antihistamine (levocabastine) in comparison with a non-FDA approved oral antihistamine (terfenadine). The other³ included a pharmacodynamic study in an environmental exposure chamber and a trial of combination therapy not identified as an intervention of interest for the present review.
• Oral selective antihistamine versus intranasal corticosteroid: Two reviews^{28, 147} found evidence favoring intranasal corticosteroid over mixed oral selective and non-selective antihistamine. One¹⁴⁹ of these included two trials that reported only physiologic outcomes (e.g., nasal airflow), and the other²⁸ included trials in PAR.
• Intranasal corticosteroid versus nasal cromolyn: The AAAAI practice parameter³ recommended intranasal corticosteroid for most patients based on a single trial¹²⁵ that was one of four trials included in the present review for this comparison.
• Combination oral selective antihistamine plus intranasal corticosteroid versus intranasal corticosteroid: One review¹⁴⁷ reported greater improvement in quality of life with combination therapy than with intranasal corticosteroid monotherapy based on one trial. This trial⁶² was one of five trials included in the present review for this comparison.

Discordant conclusions of comparable effectiveness (equivalence)

• Intranasal corticosteroid versus nasal antihistamine: All three reviews supported the use of intranasal corticosteroid based on a systematic review²⁹ that included patients with PAR.
• Intranasal corticosteroid versus oral leukotriene receptor antagonist: Two reviews^{28, 147} found evidence to support the use of intranasal corticosteroid for symptoms of AR²⁸ and SAR.¹⁴⁷ In both of these, conclusions were based on statistical rather than clinical significance of treatment effects reported by two^{128, 129} (in the ARIA guidelines²⁸) or four^126-129 (in the 2009 systematic review¹⁴⁹) of five trials^{97, 126-129} included in the present review for this comparison.
• Combination intranasal corticosteroid plus nasal antihistamine versus nasal antihistamine: One review¹⁴⁷ found evidence to support the use of combination therapy over nasal antihistamine (azelastine) monotherapy based on one¹²¹ of five^{115, 117, 121} trials included in the present review for this comparison. Four trials^{115, 117} were published after the systematic review.¹⁴⁷

Applicability

Applicability is assessed in terms of PICOTS—populations, interventions, comparisons, outcomes, timeframes, and settings of care.⁷⁹

Populations: In studies that reported the ethnic-racial make-up of trial participants, most patients were white. Approximately 9 percent were black, 7 percent were Hispanic, and 3 percent were Asian. Adult patients tended to be in their 30s and 40s. Some trials included older patients. This population is representative of many patients with SAR. Results are likely to be generalizable to adults of different ethnicities or ages, although this is not known with certainty. Patients in their seventh or eighth decade of life may require dosage adjustments for reduced renal or liver function and greater vigilance for adverse events, for example, sedating effects.

Most patients had moderate to severe symptoms at baseline and had a minimum 5-year duration of SAR. Evidence from three small trials that studied patients with mild symptoms,^{122, 128, 130} was insufficient to suggest that mild nasal symptoms respond differently than moderate/severe symptoms to the specific treatments compared. Only one¹³⁰ of these trials reported disease duration; 92 percent of patients had SAR for more than 5 years. Applicability of the findings from this report to patients with mild or recent onset SAR is therefore unknown.

Interventions/Comparisons: The interventions investigated represent treatment options for SAR currently available in the United States. We restricted our search to trials that used FDA-approved drugs at FDA-approved doses. Conclusions may not be applicable to drugs in included drug classes that were not specifically studied (e.g., drugs or doses used in Europe). Additionally, for comparisons with trials studying a small proportion of the drugs in a class, applicability of the findings to other drugs in the class that were not studied is uncertain. We did not assess eye drops for the treatment of eye symptoms associated with SAR. We sought but did not find sufficient comparative trials to address as-needed dosing.

Outcomes: SAR symptoms comprise nasal, eye, ear, palate, and throat symptoms. To maximize comparability across trials, we focused on the most often reported nasal and eye symptom outcomes, which likely enhances generalizability. For patients who experience less common symptoms, such as postnasal drip or ear itching, results from this report may not be generalizable. Outcomes were reported within the time frames of their trials and comparators; for example, nasal outcomes at 2 and 4 weeks were not mixed for trials involving intranasal corticosteroid, but were mixed for trials involving other drugs of more uniform onset and duration of action. It is unclear whether results from shorter intervals are generalizable to longer use, for example, whether treatment effectiveness reported at earlier time points is maintained at later time points and whether the incidence of adverse effects increases with increased duration of exposure.

Timeframes: By limiting diagnosis specifically to SAR, we excluded not only patients with PAR but also those classified according to the new criteria proposed by the ARIA group,²⁸ that is, those diagnosed with intermittent allergic rhinitis or persistent allergic rhinitis. These categories define overlapping but differing patient populations.³ Diagnostic categories vary in at least two dimensions: duration of allergen exposure and type of allergen. Because treatments are symptomatic, it is not expected that type of allergen will affect treatment response. However, duration of allergen exposure and, consequently, of treatment exposure may impact the applicability of the findings. For the assessment of treatment effectiveness in real-world settings, we included studies with a 2-week minimum treatment duration during pollen season.⁵⁰ Study duration was therefore limited by the natural pollen cycle. We searched for trials of longer duration to compare short-term (weeks) and longer-term (months) effectiveness and harms, but the few trials of longer than 4 weeks' duration identified prevented definitive conclusions. Similarly, patients who require less than 2 weeks' treatment may experience different effects than those reported here.

Settings: Of all trials identified, only one was not set in Europe or North America. This was a trial of 50 patients conducted in Asia (Nepal).¹¹⁶ Across all trials, Asian patients represented a minor fraction of patients studied. Generalizability to patients in Asia or to Asian patients in North America or Europe may be limited by differing aeroallergen exposure and by potential genotype differences affecting metabolism of drugs used to treat SAR.

Implications for Clinical and Policy Decisionmaking

Fortunately, SAR is not a life-threatening disease. Consideration of risks and benefits of treatment therefore shifts, from an expectation that adverse events may accompany effective treatments to an appreciation that adverse effects of treatment may be worse than the disease itself. We did not find high strength evidence for differences in effectiveness or adverse effects in any treatment comparison. We did find high strength and moderate strength evidence for comparable effectiveness of several treatments for several outcomes, low strength evidence for superiority of two treatments for two outcomes, and moderate strength evidence for the avoidance of insomnia.

This evidence may be insufficient for policy decisionmaking. For example, although conclusions of comparable effectiveness may suggest that differential costs of treatments are unwarranted, lack of evidence to evaluate comparative harms of these treatments prohibits full assessment of their risk-benefit profiles.

For clinical decisionmaking, conclusions of comparable effectiveness suggest that patient preferences and priorities can contribute significantly to treatment choice. When considering the balance between effectiveness and harms in relatively healthy individuals, potential harms may acquire greater weight.¹⁴⁴

Limitations of the Comparative Effectiveness Review Process

To narrow the scope of this project to a manageable size, we made several decisions at the start that had downstream consequences. These included:

• We restricted diagnosis to SAR. Given the current state of transition between classification schemes for allergic rhinitis, use of the original scheme may have excluded some trials. However, it is acknowledged that SAR and intermittent allergic rhinitis define different patient populations.³ We decided to pick one disease to study and then find studies similar enough to compare results. Introducing studies of allergic rhinitis classified according to the newer scheme may have added to the variability of included studies.
• We did not examine every possible treatment comparison. Rather, guided by input from Key Informants and the TEP, we prioritized comparisons that reflect treatment decisions encountered in the clinical setting. It is hoped that we selected, and found evidence to assess, comparisons that are meaningful to users of this report.
• We excluded trials of one drug versus a placebo and focused on direct comparisons only. This decision was based on feasibility concerns given the large scope of the project and time constraints. Harms assessment was limited by the absence of placebo groups, which can inform adverse event reporting particularly.
• We included FDA-approved drugs only. For the comparison of oral selective antihistamine to oral nonselective antihistamine in particular, this significantly reduced the number of included trials. The majority of these trials used terfenadine or astemizole as the selective antihistamine comparator, neither of which is currently FDA-approved due to postmarketing safety concerns. As a result, only three trials were included for this comparison.
• Our minimum 2-week duration excluded examination of other treatment features that may be important to patients, e.g., onset of action and harms associated with shorter exposure. However, harms associated with the interventions as defined (i.e., minimum 2-week exposure) were included. Trials less than 2 weeks' duration often did not replicate natural methods of exposure to airborne allergens (i.e., used instead environmental exposure chambers, direct application of allergen, or prolonged weekend visits to parks), and results may be less applicable.
• As described below, reporting of efficacy outcomes in SAR research currently is nonstandard. To maximize our ability to compare outcomes across trials, we selected the most commonly used symptom measures, namely the four-symptom TNSS and the three-item total ocular symptom score (TOSS). Data from trials that used variations on these reporting scales could not be incorporated into the report. Symptoms potentially important to patients but seldom assessed (e.g., post-nasal drip, and ear and palate itching) were not included in this review.
• The scope of this report is class comparisons of SAR treatments. As a consequence of this approach, individual drug comparisons were beyond the scope of this report. Also, for comparisons with trials studying a small proportion of the drugs in a class, we were limited in our ability to make conclusions about entire pharmacologic classes, particularly for larger classes such as intranasal corticosteroids and oral nonselective antihistamines. As discussed above (in the Discussion of KQ1), the impact of this limitation may be small for certain drug classes, such as oral nonselective antihistamine, which are less commonly used, and oral decongestant, of which the more commonly used drug was studied.
• Limitations in the quality of trial reporting impacted directly the conclusions that could be drawn and strength of evidence ratings. For example, insufficient group-level data reporting prevented equivalence assessments. Insufficient descriptions of analyses compromised quality ratings. For example, if intention-to-treat (ITT) analysis was not specified, or insufficient patient flow data were provided to determine that an ITT analysis was done, trial quality was rated poor. This was a conservative decision that was warranted. It is hoped that continued implementation of guidelines for trial reporting will address such difficulties.

Limitations of Evidence Base

In their review of SAR treatments, Benninger et al. (2010) conclude that “the reporting of published data should be standardized to permit comparisons among treatments.”⁴¹ The following six improvements are cited:

• Standard inclusion criteria for allergic rhinitis based on a unified definition
• Standard stratification of disease severity
• TNSS based on four nasal symptoms (congestion, rhinorrhea, sneezing, and nasal itch) and reported on a 3-point or 4-point severity scale
• Standard ocular data for TOSS
• Standard quality of life data using a validated survey
• Standard age cutoffs (adult studies should include ages > 18 years; adolescents, 12–18 years old; school-age children, 6–11 years old)

In our experience, factors 2, 3, 4, and 6 reduced our ability to draw evidence-based conclusions. Additionally, evidence for efficacy and harms in patients with mild disease was lacking due to enrollment of patients primarily with moderate/severe disease. In fact, standard definitions of mild, moderate, and severe disease in terms of symptom scales do not currently exist. A 4-point scale may be divided into terciles (0-1 mild, 1-2 moderate, 2-3 severe), but this is an empirical division. It is unknown whether the scale is linear in patients' experience.

We could not incorporate several trials that reported only total symptom scores, comprising nasal, eye, ear, and palate symptoms. Standard reporting of the four-symptom TNSS and the three-symptom TOSS would greatly facilitate treatment comparisons. It is surprising that the list above does not include a call for well-defined MCIDs for symptom scales. Although our selection of clinically informed MCIDs permitted us to draw clinically relevant conclusions, validation of the values used (30 percent maximum score) using anchor-based approaches is desirable. Without such well-defined MCIDs, at least three analytic tools important for clinical research–power calculations, non-inferiority margins, and responder analyses–are compromised.

Another methodological issue is the incomplete reporting of results. Examples include:

• Reporting the results of statistical testing for only two arms of a three-arm trial
• Not reporting variance estimates for group-level treatment effects
• Not reporting results for all identified outcomes
• Missing baseline symptom or quality of life scores
• Partial accounting of patient flow through the trial

Adverse event reporting was consistently incomplete. Severity of adverse events was sometimes mentioned, but, as above, lack of standard definitions of severity or a standard adverse event scale currently limits the usefulness of severity descriptions. Adverse events often were not reported by treatment group or were not identified. That is, the proportion of patients experiencing adverse events was at times reported without any description of the adverse events experienced.

We excluded several trials that did not report results by age groups or that formed age groups using non-standard cut points. Defining “adolescent” from age 12 may be arbitrary, but its general adoption would permit greater learning about this age group. FDA commonly uses a cut point of age 12 for dosing of SAR drugs in children and adolescents.

Finally, within the constraints of our inclusion criteria, we identified few to no studies for the assessment of SAR treatments in pregnant women and children. Head-to-head active comparator trials may be ethically difficult in these vulnerable populations unless true equipoise exists. Although we preferred RCTs, we would have included relevant nonrandomized comparative trials, but we found none.

Research Gaps

The greatest need in SAR research is increased methodological rigor. Widely used symptom rating scales require standardization and validation. Lack of anchor-based MCIDs is a major deficiency. Agreed-upon reporting standards for effectiveness and harms outcomes are needed. Agreed-upon classifications of patients by age and standardized definitions of symptom and harms severity also are needed. Study designs that can more efficiently assess the effects of additive therapies are lacking. That is, studies in which all patients are treated with one component of a combination (e.g., oral selective antihistamine) and only those who are resistant receive the second component (e.g., intranasal corticosteroid) may more efficiently isolate the additive effect of the second component. We identified one trial with this design.¹⁴⁸

Lack of evidence on populations of interest is a research gap. Currently, the majority of trial participants are relatively homogenous: white and middle-aged with moderate/severe SAR symptoms. Inclusion of different races, greater proportions of patients toward both ends of the age spectrum, and patients with mild symptoms may inform our understanding not only of the comparative effectiveness and harms of SAR treatments in different groups, but also of the expression of SAR in various ethnic groups, the natural history of the disease across the life span, and the effect (if any) of early treatment on later symptom expression. As noted above, however, ethical considerations may limit the inclusion of vulnerable populations (e.g., children) in well-designed studies of pharmacologic interventions.

For pregnant women, pregnancy registries and rigorous studies based on the data therein can fill the gap. Additionally, greater understanding of how the physiologic changes of pregnancy affect the magnitude and direction of change in drug disposition may facilitate application of effectiveness and safety findings from the nonpregnant population to pregnant women. This presumes use of Pregnancy Category B drugs to avoid potential known or unknown teratogenic effects of other drugs.

Conclusions

For most treatment comparisons of interest, evidence was insufficient to support conclusions about comparative effectiveness and harms. Of conclusions that could be drawn, most suggested comparable effectiveness of treatments compared. For adults and adolescents over the age of 12 we found:

• High strength evidence for comparable effectiveness (equivalence) of:
  • Combination intranasal corticosteroid (fluticasone propionate) plus nasal antihistamine (azelastine), intranasal corticosteroid monotherapy, and nasal antihistamine monotherapy for nasal and eye symptoms at 2 weeks.
  • Intranasal corticosteroid and oral leukotriene receptor antagonist (montelukast) for nasal symptoms at 2 weeks.
• Moderate strength evidence for comparable effectiveness of oral selective antihistamine and oral leukotriene receptor antagonist for nasal and eye symptoms and for improved quality of life at 2-4 weeks.

In this population, we found evidence for the superiority of:

• Oral selective antihistamine over both oral decongestant and combination oral selective antihistamine plus oral decongestant to avoid insomnia at approximately 2 weeks (moderate strength evidence).
• Oral leukotriene receptor antagonist over oral selective antihistamine for reduced asthma rescue medication use at 2-4 weeks (moderate strength evidence).
• Combination oral selective antihistamine plus intranasal corticosteroid over oral selective antihistamine monotherapy for improved quality of life at 2-4 weeks (low strength evidence).

Conclusions about symptom improvement were based on clinically informed but non-validated MCIDs. Sensitivity analyses supported the conclusions above as well as the use of combination oral selective antihistamine plus oral decongestant over oral selective antihistamine monotherapy for nasal symptoms. The lack of comparative evidence for all drugs within each class limited the applicability of conclusions. Evidence was insufficient or lacking to support any of 48 other identified treatment comparisons of interest among adults and adolescents over the age of 12, pregnant women, and children younger than 12 years of age.