Claustrophobia during magnetic resonance imaging: Cohort study in over 55,000 patients


  • Results were presented in part as an abstract at the RSNA 2007 meeting (no. 22-09SSC).



To evaluate whether MR scanners with acoustic noise reduction and a short magnetic bore reduce the rate of claustrophobic reactions.

Materials and Methods

We performed a cohort study in an outpatient setting, enrolling a total of 55,734 consecutive patients referred for MRI of any part of the body based on a clinical indication. Imaging was performed using a conventional MR scanner (42,998 patients) and a recently developed MR scanner (12,736 patients) with 97% acoustic noise reduction and a short bore. Multiple logistic regression analysis was used to adjust for the nonrandomized design.


In addition to those undergoing head-first examinations, female and middle-aged patients were significantly more likely to develop claustrophobia in the logistic regression analysis (P < 0.001). The rate of claustrophobic reactions was significantly lower with the recent MR scanner (0.7%; 95% confidence interval [CI]: 0.6–0.9%) than with the conventional scanner (2.1%; 95% CI, 2.0–2.3%; P < 0.001) with an adjusted odds ratio (OR) of 3.1 (95% CI, 2.5–3.9) and a number needed to treat of 72 (95% CI, 63–85).


The incidence of claustrophobia may be reduced by a factor of 3 when recently-developed MR scanners are used. J. Magn. Reson. Imaging 2007;26:1322–1327. © 2007 Wiley-Liss, Inc.

MAGNETIC RESONANCE IMAGING (MRI) has been described as the most important medical innovation in the last 25 years (1). Between 1% and 15% of all patients who undergo an MR examination suffer from claustrophobia and cannot be imaged, or they require sedation to complete the scan (mean, 2.3%; 95% confidence interval [CI], 2.0–2.5%; Table 1) (2–12). Claustrophobic patients are frightened and experience a feeling of confinement or being closed in (13, 14). In these patients anxiolytic sedation and additional sequences (after sedation) may be necessary to complete the examinations. This situation reduces workflow, limits patient acceptance, and wastes valuable scanning time. Improved convenience during the MR examination (e.g., reduction of noise and sensations of confinement) appears to be essential to avoid claustrophobic reactions (6, 15). Thus, we have performed an observational cohort study to evaluate whether a newly-designed MR scanner with acoustic noise reduction and a patient-friendly short magnetic bore reduces the rate of claustrophobic reactions.

Table 1. Rate of Claustrophobia in Previous Clinical MR Imaging Studies and the Present Study*
StudyYear of publicationNo. of patients includedClaustrophobia, N (%)a
  • *

    Results are provided as absolute numbers with percentages and 95% CIs. Please note that the sum of all previous studies is smaller (by more than a factor of 3) than that of the present study. Values in bold represent the averages of all previous studies and the ratios of claustrophobia in the conventional and recently developed MR scanner.

  • a

    Values in square brackets are 95% CIs. The CIs were estimated using the score method as described (19).

  • b

    Summary values for all previous studies are not weighted according to study size.

Weinreb et al (12)19844503 (0.7% [0.2–1.9])
Hricak and Amparo (8)1984116031 (2.7% [1.9–3.8])
Brennan et al (3)1988524 (7.7% [3.0–18.2])
Flaherty and Hoskinson (6)19892109 (4.3% [2.3–7.9])
Avrahami (2)1990300046 (1.5% [1.2–2.0])
Kilborn and Labbe (30)19901087 (6.5% [3.2–12.8])
Dantendorfer et al (5)1991568250 (0.9% [0.7–1.2])
MacKenzie et al (31)19955005 (1.0% [0.4–2.3])
Murphy and Brunberg (9)1997939134 (14.3% [12.2–16.7])
Sarji et al (11)1998332418 (0.5% [0.3–0.9])
Francis and Pennell (7)2000175474 (4.2% [3.4–5.3])
Kaltenthaler et al (28)200478224 (3.1% [2.2–4.9])
All previous studiesb 17,961405 (2.3% [2.0–2.5])
Present study (conventional MR) 42,998911 (2.1% [2.0–2.3])
Present study (recent MR) 12,73693 (0.7% [0.6–0.9])
Present study (all examinations) 55,7341004 (1.8% [1.7–1.9])


Study Design

We performed a cohort study in an outpatient setting, enrolling a total of 55,734 consecutive patients referred for MRI over a period of over eight years. Between 28 April 2004 and 23 December 2005 we analyzed claustrophobia in 12,736 consecutive patients undergoing imaging in an outpatient imaging center ( on a 1.5T MR scanner (Magnetom Avanto; Siemens Medical Solutions, Erlangen, Germany) with 97% acoustic noise reduction (below 99 dB A) and a short (1.6 m) and wide conical-shaped magnetic bore (0.6 m) that resembles the appearance of the gantry of a computed tomography scanner (Fig. 1). The results obtained in this patient group were compared with those achieved in a consecutive cohort (42,998 patients) examined on a conventional 1.0T MR scanner (Magnetom Impact Expert Plus; Siemens) in the same center between 3 November 1997 and 3 April 2004. This conventional MR scanner had a flat front and a bore with a length of 2.25 m and a width of 0.6 by 0.53 m (Fig. 1) with an acoustic noise of up to 128 dB A. All other parameters (e.g., receptionists with whom patients interact, dressing rooms, waiting rooms, etc.) were kept constant during the entire study period. Hearing protection was used in all patients as required by federal law. The study was approved by the State Ethics Committee. Reasons other than claustrophobia to prevent MRI were analyzed in a separate report (16). After verbal informed consent was obtained (written informed consent was waived by the State Ethics Committee because of the nonexperimental design of the study) patients were included in this cohort study. All procedures were conducted according to the Declaration of Helsinki.

Figure 1.

Appearance of the clinical MR scanners used. a: Conventional MR scanner with a flat front and a bore with a length of 2.25 m and a width of 0.6 by 0.53 m. b: Recent MR scanner with a short (1.6 m) and wider conical-shaped magnetic bore (0.6 m).

Data Analysis

In all 55,734 consecutive patients, claustrophobia was determined to be present by one physician if an anxious or panic reaction occurred that: 1) required intravenous sedation with benzodiazepines (2) (10 mg diazepam, Ratiopharm, Ulm, Germany); or 2) prevented MRI because the patient refused sedation (combined end-point). Sedation with diazepam was administered according to the standards of the American Society of Anesthesiology (17) and was assessed as an event since patients can suffer rare but serious injury (e.g., severe motor vehicle accidents) after being discharged (18). If patients underwent MR more than once on the same scanner, only the first examination was included in the statistical analysis. Patient characteristics and data on claustrophobic reactions were collected anonymously using radiology information software (RadCentre 6; iSOFT, Manchester, UK).

Secondary Outcomes

We performed an intraindividual comparison of the rate of claustrophobic reactions among patients who underwent MR examinations of the same body region with both scanners and compared patients who were examined before and after installation of the recent scanner (years 2003 and 2004) to account for a potential time bias. The success rate of intravenous benzodiazepine sedation was also determined.

Statistical Analysis

The chi-squared and unpaired t-test were used as appropriate for categorical and continuous variables. Rates of claustrophobic reactions and corresponding 95% CIs were obtained using the score method as described (19). Differences between rates were examined with chi-squared tests and CIs for differences were also calculated with the score method (20). Odds ratios (ORs) and the number needed-to-treat were calculated to determine the clinical impact of the recent scanner. Their CIs were computed with the logit method as described (21). Multiple logistic regression was used to adjust for the nonrandomized design. For intraindividual comparison of the rates of claustrophobia the McNemar test was used and CIs were calculated by the score method for paired rates (22). Statistical analyses were conducted using SPSS version 12.0 and Confidence Interval Analysis for Windows version 2.1.2. P values of 0.05 or less were considered significant.


Table 2 lists the characteristics of the 55,734 patients. Head-first MR examinations, which increase claustrophobic stress for patients, were significantly more common with the recent scanner (P < 0.001), mainly because of a higher rate of abdominal and pelvic examinations (Table 2). In addition to those undergoing head-first examinations, female and middle-aged patients were significantly more likely to develop claustrophobia (P < 0.001; Tables 3 and 4; Fig. 2).

Table 2. Characteristics of the Patients and the MR Examinations by Type of Scanner Used (N = 55,734)
CharacteristicConventional MR scanner (N = 42,998)Recent MR scanner (N = 12,736)P valuea
  • a

    P values were obtained using the chi-squared and t-test. Results of the logistic regression analysis for the comparison of the two MR scanners are provided in Table 4.

  • b

    The higher age of patients coincides with an increase during the study period of 5.3 years in the mean age of inhabitants of the rural district in which the imaging center is located, according to data provided by the State Statistical Department.

Age (year)47.0 ± 17.751.0 ± 17.5<0.001b
Gender, N (%)  0.110
 Male20,527 (47.7)6184 (48.6) 
 Female22,471 (52.3)6552 (51.4) 
Type of health insurance, N (%)  <0.001
 State health insurance40,499 (94.2)12,175 (95.6) 
 Private health insurance1302 (3.0)295 (2.3) 
 Trauma insurance1197 (2.8)266 (2.1) 
Time of MR examination, N (%)  <0.001
 Mornings15,399 (35.8)5646 (44.3) 
 Afternoons27,599 (64.2)7090 (55.7) 
Scanning direction, N (%)  <0.001
 Head-first MR examinations35,087 (81.6)10,911 (85.7) 
 Feet-first MR examinations7911 (18.4)1825 (14.3) 
Region of MR examination, N (%)  <0.001
 Head and neck19,287 (44.9)5391 (42.3) 
 Thorax493 (1.1)167 (1.3) 
 Upper extremity3171 (7.4)977 (7.7) 
 Abdomen/pelvis12,133 (28.2)4353 (34.2) 
 Lower extremity7914 (18.4)1848 (14.5) 
Table 3. Rate of Claustrophobia in Subgroups of Patients in the Unadjusted Analysis (N = 55,734)*
CharacteristicN/N%P valuea
  • *

    Values in parentheses are 95% CIs. The confidence intervals were estimated using the score method as described (20).

  • a

    P values were obtained using the chi-squared test (unadjusted analysis). All characteristics were used as input for the logistic regression analysis (adjusting for the nonrandomized design), which is presented in Table 4.

Age  <0.001
 Below 40 years250/18,6281.3% (1.2–1.5) 
 40 to 65 years482/18,6092.6% (2.4–2.8) 
 Above 65 years272/18,4971.5% (1.3–1.7) 
Gender  <0.001
 Male350/26,7111.3% (1.2–1.5) 
 Female654/29,0232.3% (2.1–2.4) 
Type of health insurance  <0.001
 State health insurance959/52,6741.8% (1.7–1.9) 
 Private health insurance24/15971.5% (1.0–2.2) 
 Trauma insurance21/14631.4% (0.9–2.2) 
Time of MR examination  <0.001
 Mornings454/21,0452.2% (2.0–2.4) 
 Afternoons550/34,6891.6% (1.5–1.7) 
Scanning direction  <0.001
 Head-first MR examinations986/45,9982.1% (2.0–2.3) 
 Feet-first MR examinations18/97360.2% (0.1–0.3) 
Region of MR examination  <0.001
 Head and neck558/24,6782.3% (2.1–2.5) 
 Thorax22/6603.3% (2.2–5.0) 
 Upper extremity79/41481.9% (1.5–2.4) 
 Abdomen/pelvis327/16,4862.0% (1.8–2.2) 
 Lower extremity18/97620.2% (0.1–0.3) 
Table 4. Significant Results of the Logistic Regression Analysis (N = 55,734)*
CharacteristicAdjusted OR95% CI
  • *

    CIs for adjusted ORs were computed with the logit method as described (21). CIs with the lower limit above 1.0 indicate significant influence of the characteristic. Middle age was defined as 40–65 years.

Conventional scanner vs. recent scanner3.12.5–3.9
Middle age vs. all other ages2.11.8–2.4
Women vs. men1.71.5–1.9
Head-first vs. feet-first MR examinations10.66.6–16.9
Figure 2.

Frequency of claustrophobia in different age groups. Absolute values are presented for patients without (colored blue) and with claustrophobia (colored red). Middle-age patients (40–65 years) were more likely to develop claustrophobia. See Table for the detailed statistical analysis (logistic regression) of the effect of age on claustrophobia.


The rate of claustrophobic reactions was significantly lower among the patients examined on the recent scanner (93 of 12,736 patients; 0.7%; 95% CI, 0.6–0.9%; Table 1) than among the patients examined on the conventional MR scanner (911 of 42,998 patients; 2.1%; 95% CI, 2.0–2.3%; P < 0.001). The absolute difference between the claustrophobia rates on the two scanners was 1.4% (95% CI, 1.2–1.6%). The corresponding unadjusted and adjusted ORs were 2.9 (95% CI, 2.4–3.6) and 3.1 (95% CI, 2.5–3.9; Fig. 3; Table 4), respectively, and 72 patients would have to undergo MRI with the recent scanner to prevent one event (number needed-to-treat; 95% CI, 63–85).

Figure 3.

ORs for the conventional vs. the recent MR scanner. ORs (with 95% CIs) comparing both MR scanners in terms of all claustrophobic reactions (top), of claustrophobic reactions preventing MR imaging (middle), and requiring intravenous sedation (bottom). All ORs are higher than 1 and thus indicate higher risk in conventional scanners. Closed squares are unadjusted ORs and open yellow squares are adjusted ORs. Adjusted ORs are always higher than unadjusted ORs, further increasing the reliability of the conclusions. Please note that the size of the squares indicates the reciprocal of the variance (i.e., larger squares as for reactions preventing MR imaging (middle) are equal to lower variance).

MRI was less frequently prevented by claustrophobia in the case of the recent scanner (57 of 12,736 patients; 0.4%) than with the conventional MR scanner (337 of 42,998 patients; 0.8%; P < 0.001; OR, 1.8; 95% CI, 1.3–2.3; Fig. 3). Also, the need to administer intravenous benzodiazepines was lower with the recent scanner (36 of 12,736 patients, 0.3%) than with the conventional scanner (574 of 42 998 patients; 1.3%; P < 0.001; OR, 4.8; 95% CI, 3.4–6.7; Fig. 3).

Secondary Outcomes

In the intraindividual analysis of 2630 patients who underwent MR imaging with both scanners, the rate of claustrophobic reactions was again significantly lower with the short-bore scanner (18 of 2630 patients; 0.7%; 95% CI, 0.4–1.1%) than the conventional scanner (59 of 2630 patients; 2.2%; 95% CI, 1.7–2.9%; P < 0.001). For patients who were examined before and after installation of the recent scanner (years 2003 and 2004), the adjusted OR was 3.2 (95% CI, 2.1 to 4.8; P < 0.001). Of the 610 patients who received intravenous sedation with benzodiazepines prior to imaging, 608 patients were able to undergo the MR examination (99.7% success rate; 95% CI, 98.8–100%).


Claustrophobic reactions are a major obstacle to performing MR examinations. Our large cohort study indicates that incidence of claustrophobia can be reduced by a factor of 3 when patients are imaged on an MR scanner with a conical-shaped short magnetic bore and reduced acoustic noise.

Clinical Implications

The study also indicates that recently developed scanners may lower costs. Three patients are scheduled for imaging per one hour in the imaging center (45–50 patients per day), with each imaging slot being worth approximately $250 (including all purchase and maintenance costs). Because of the high throughput, no prescreening for claustrophobia can be performed. Consequently, $250 in reimbursement is lost if one examination has to be cancelled due to claustrophobia. With a throughput of 45–50 patients per day in the imaging center, claustrophobia on the recent scanner occurs every third day, which is relevantly reduced in comparison to examinations on the conventional scanner (one patient per day). Of course, the decision to purchase a new MR scanner depends not only on the potential to reduce claustrophobia but on many other factors such as image quality. It should be noted that the recent short-bore MR scanner not only retains the field (1.0T) and gradient strength (20 mT/m) of the conventional scanner but extends these values to 1.5T and 45 mT/m and thereby improves image generation. More important, sedation can result in adverse events (approximately 0.4–2.9%) (23, 24) and also severe injury and malpractice litigation for compensation of injured patients (18) and should be avoided whenever possible. The intraindividual comparison also showed that claustrophobia was significantly less likely to occur in patients imaged on the short-bore scanner. The reason for the improvement is most likely the patient-centered design (short conical-shaped bore), which limits apprehension caused by the environment of the MR scanner (14) and adheres to the strategy of “human factors engineering” described in the “Quality System Regulation” of the U.S. Food and Drug Administration (

Our study confirms the results of a smaller study (50 patients) reporting a reduction in claustrophobia with a patient-friendly scanner design (25). Such so-called “open scanners” have also been advocated for MRI but have less powerful magnets and are inferior in terms of diagnostic performance. A simple well-established maneuver to alleviate claustrophobia is turning patients prone during the MR examination to allow them to see the opening of the magnet, but this maneuver cannot be performed in all patients because of placing issues (8). Oral benzodiazepines, prism glasses, communication devices (26), having a relative or friend present in the room, and music (27) are other long-established options to reduce claustrophobic responses to MR examinations. Prism glasses, a communication device, hearing protection, and having a friend or relative present were available with both MR scanners used in our study. Since the rate of claustrophobia in the 42,998 patients examined on the conventional MR scanner in our study is smaller than that reported in most previous publications (Table 1) (3, 4, 6–10, 28) it is likely that the improvement in claustrophobia by a factor of 3 that we observed with the recent MR scanner is not overestimated. The sedation in our study was performed according to the guidelines of the American Society of Anesthesiology (17) by a physician in an outpatient setting with a high success rate; however, nurse-led sedation services for MR examinations might increase the quality of sedation in hospitals and reduce cost (23, 29).

A claustrophobic rate as small as that present in the 12,736 patients examined with the recent short-bore MR scanner (0.7%) was also found in a study on 3324 Malaysian patients examined with a conventional MR scanner (Table 1; 0.54%) (11). This study suggested the following risk factors for the development of claustrophobia: middle age, male sex, and a higher socioeconomic status and education level. The authors speculate that the lower education level is one of the reasons for a lower claustrophobia rate in their study when compared with those conducted in Western Europe and North America. Also, cultural difference might play a role in differences found in claustrophobia rates. In contrast to this smaller study, we found that women were significantly more likely than men to experience claustrophobia during MRI, whereas an age between 40 and 65 years was confirmed as a risk factor in our cohort.

Study Limitations

Our study is by a factor of 3 larger than the sum of other studies published on claustrophobia during MRI (Table 1). Nevertheless, it is important to note that our findings are limited by the fact that the patients were not prospectively randomized to be imaged using one of the two MR scanners. Due to the consecutive nature of the trial and the inclusion of all patients, the cohort is representative of patients undergoing an outpatient MR scan. Also, we cannot rule out the possibility that a before-after bias might have influenced the present study. However, we adjusted for confounders and different distributions in both groups with logistic regression, and the adjusted ORs favoring the recent MR scanner were even higher than the unadjusted ORs. Furthermore, given the sequential sampling design of the study, it is conceivable that a referral bias could have influenced the results; this effect was extensively addressed in our secondary analyses. Finally, characteristics such as patient weight and height were not available from the patients' electronic health card used at the time the trial was conducted. Psychological claustrophobia severity, image quality, and examination duration evaluation was likewise not performed. To analyze claustrophobia in a large cohort like this would be impractical using a randomized design; however, small randomized studies might provide further insights into claustrophobic reactions. Given that despite the higher rate of head-first examinations, there was significantly less claustrophobia on the short-bore scanner and because of the similar results in the intraindividual substudy, we believe our findings are valid.

In conclusion, the present study compared claustrophobic reactions in a conventional and a recent short-bore MR scanner in a consecutive cohort of patients scheduled to undergo MRI and found that the incidence of claustrophobia can be reduced by a factor of 3 by using MR scanners with a patient-centered design. This might further increase the clinical utility and applicability of MRI.