**2. Method**

The preferred reporting items for systematic reviews and meta-analysis (PRISMA) checklist was used to design the study and to report results based on best practices in the field of literature reviews and meta-analyses [42].

#### **2.1 Literature research**

We conducted a literature review in these databases: PsycNet, MedLine, ProQuest, Web of Knowledge, Cochrane, and Psychology and Behavioral Sciences Collection. Search terms used in PsycNet were "panic disorder" OR "panic attack" OR panic AND sleep OR "sleep disorders" OR insomnia OR "nocturnal panic" OR "sleep deprivation." Search terms used in MedLine, Web of Knowledge, and Cochrane were panic OR "panic disorder" AND sleep OR "sleep initiation and maintenance disorders" OR dyssomnia OR "sleep deprivation." Publication date was 1980 (DSM-III year of publication) to May 2016 inclusive.

To retrieve gray literature (i.e., unpublished work or studies that are published outside widely available journals) [43], reference lists of selected articles were searched for potentially eligible articles regarding sleep in PD. The Laval University library was searched for book chapters addressing sleep in PD. Reference lists were screened for articles of interest. Inclusion of ProQuest and Web of Knowledge databases helped retrieve gray literature, as these sources contain theses, symposia, and convention papers, in addition to journal articles.

#### **2.2 Article selection**

The following inclusion criteria were used: (a) studies had to be published in English, French, or Spanish; (b) minimum participant age was 18 years; (c) to ensure population representativeness, studies including participants with comorbidities were accepted if PD was the primary diagnosis for at least one subgroup; (d) studies had to report quantitative group data for at least one sleep variable; (e) data on sleep disturbances had to include means, standard deviations, and group sizes for PD and control groups; (f) prevalence of NPA had to be reported in percentage or number of participants, and total sample size had to be included; (g) data had to be issued from self-report questionnaires, clinician-administered interviews, sleep diaries, polysomnography, or actigraphy; and (h) for subjective and objective sleep data, patients with PD had to be compared to a group of healthy controls (for the prevalence of nocturnal panic, no comparison group was necessary).

Studies that met the following criterion were excluded: (a) studies in which all patients with PD reported a physical disease or one of the following comorbid psychiatric disorders, schizophrenia, bipolar disorder, and alcohol or drug abuse or dependence, and (b) studies in which no data could be grouped using the following procedure:

Studies were classified and grouped on the basis of the measure they reported. For example, all studies reporting results on the Pittsburgh sleep quality index (PSQI) were grouped together, the same was done for the Hamilton rating scale (HAM-D), polysomnography, etc. For articles reporting NPA, grouping was carried out on the basis of the frequency criteria used. For example, all studies reporting the prevalence of patients who had at least one NPA in their lifetime were grouped together, all articles reporting the prevalence of patients having at least one NPA/month were grouped together, etc. Some articles reported data from specific sleep measures or NPA frequency criteria that were not repeated elsewhere.

**99**

**Table 1.**

*Inter-rater agreement.*

*A Meta-Analysis of Sleep Disturbances in Panic Disorder DOI: http://dx.doi.org/10.5772/intechopen.86306*

graphic/actigraphic data, and prevalence of NPA).

additional study were excluded.

**2.3 Data extraction**

**2.4 Inter-rater agreement**

*2.4.1 Article selection*

*2.4.2 Data extraction*

**Study description**

**Self-report data**

**NPA**

**Polysomnographic data**

*Note: NPA = nocturnal panic attack; PD = panic disorder.*

Studies reporting data that could not be grouped with other data from at least one

Once reference retrieval was complete, each of the selected papers was reviewed,

To reduce selection bias, a second trained judge independently rated the eligibility of 30% of the retrieved literature. The results of the two judges' ratings were compared, and cases of disagreement were discussed to reach consensus.

Five raters participated in data extraction. Four of the judges were undergraduate psychology students, and one was a graduate psychology student (AP). A pilot coding was conducted, wherein all judges used the coding manual to rate the same

**Item Percentage inter-rater agreement**

PD group sample size 100.0 Control group sample size 100.0 Mean age 88.2 Gender 73.1 Medication permitted or not 96.2 Percentage of participants taking medication 84.6 Psychometric instrument used to diagnose PD 82.7 Psychometric instrument used to assess sleep 100.0

Sleep variable—PD group 88.9 Sleep variable—control group 88.9

Sleep variable—PD group 100.0 Sleep variable—control group 98.8

Percentage of sample reporting NPA 46.7 NPA frequency criterion 66.7

and data was extracted using a standardized rating form and coding manual. Extracted variables included information about the sample and sampling method, general methodology, and sleep characteristics (self-report sleep data, polysomnoStudies reporting data that could not be grouped with other data from at least one additional study were excluded.
