Percutaneous laser disc decompression versus conventional microdiscectomy for patients with sciatica: Two-year results of a randomised controlled trial

Abstract

Background

Percutaneous laser disc decompression is a minimally invasive treatment, for lumbar disc herniation and might serve as an alternative to surgical management of sciatica. In a randomised trial with two-year follow-up we assessed the clinical effectiveness of percutaneous laser disc decompression compared to conventional surgery.

Materials and methods

This multicentre randomised prospective trial with a non-inferiority design, was carried out according to an intent-to-treat protocol with full institutional review board approval. One hundred and fifteen eligible surgical candidates, with sciatica from a disc herniation smaller than one-third of the spinal canal, were randomly allocated to percutaneous laser disc decompression (n = 55) or conventional surgery (n = 57). The main outcome measures for this trial were the Roland-Morris Disability Questionnaire for sciatica, visual analogue scores for back and leg pain and the patient's report of perceived recovery.

Results

The primary outcome measures showed no significant difference or clinically relevant difference between the two groups at two-year follow-up. The re-operation rate was 21% in the surgery group, which is relatively high, and with an even higher 52% in the percutaneous laser disc decompression group.

Conclusion

At two-year follow-up, a strategy of percutaneous laser disc decompression, followed by surgery if needed, resulted in non-inferior outcomes compared to a strategy of microdiscectomy. Although the rate of reoperation in the percutaneous laser disc decompression group was higher than expected, surgery could be avoided in 48% of those patients that were originally candidates for surgery. Percutaneous laser disc decompression, as a non-surgical method, could have a place in the treatment arsenal of sciatica caused by contained herniated discs.

Introduction

The prevalence of sciatica due to lumbar disc herniation is high and ranges between 1.2% to 43% depending on the studies cited.¹ Each year 5–10 per 1000 inhabitants develop sciatica in Western countries.² Natural history is generally favourable, with diminishing leg pain in six months for most patients and recovery is observed in 70–80% of patients at one to two-year follow-up.^3,4 Surgical removal of the disc herniation is the treatment of choice for refractory cases.⁴ Surgery does, however, carry the risk of complications ranging from 8–15.7%.^5,6 It is suggested that minimally invasive therapies will be able to achieve a lower complication rate with similar outcome and superior cost effectiveness.⁷ One of the minimally invasive therapies is percutaneous laser disc decompression (PLDD), a nonsurgical technique using laser energy to vaporise the nucleus pulposus via a percutaneous route.⁸ This technique has recently been evaluated for the outcome in comparison to conventional microdiscectomy.⁹ In that study it was shown that the clinical outcome at one-year follow-up, with a strategy of PLDD, was non-inferior to conventional microdiscectomy, at the cost of a longer time to recovery and higher re-intervention rates.¹⁰ Since the long-term benefit of PLDD treatment compared to microdiscectomy is unknown, we performed a two-year analysis of the study groups.

Materials and methods

We conducted a randomised controlled trial comparing PLDD with conventional surgery in a parallel group design. Eligible patients were suffering from sciatica due to a contained lumbar disc herniation refractory to conservative treatment for more than six weeks. An extensive description of the trial design was published previously.⁹

Patients and randomisation

The study was carried out in a multicentre setting entailing two university hospitals and six teaching hospitals. If a neurologist considered a patient, with sciatica that was refractory to 6–8 weeks of conservative management, to be a candidate for surgical discectomy then the patient was referred to a neurosurgeon. All patients between 18–70 years of age, were considered for inclusion in the trial if a disc herniation at the corresponding level was shown by magnetic resonance imaging (MRI) and if the herniated fragment was smaller than one-third of the spinal canal (Figure 1). This subgroup was considered suitable for this kind of treatment based on early PLDD publications.^11,12 Exclusion criteria entailed previous surgery at the same disc level, cauda equina syndrome, spondylolytic or degenerative changes, central spinal canal stenosis, pregnancy and factors preventing adequate follow-up. All eligible patients were subsequently examined and questioned by the treating neurosurgeon and a research nurse who recorded the baseline variables, follow-up questionnaires and outcome measures. Patients were randomly allocated on a 1:1 ratio to either a strategy of PLDD or conventional surgery. A computer generated randomisation list was prepared for each research nurse and each of the participating academic hospitals (n = 2) and teaching hospitals (n = 6). Blocks of random size (varying between 2–4 to minimise predictability) of random numbers were formed to ensure equal distribution of the randomisation among hospitals as well as nurses. The data manager at the Department of Medical Statistics and BioInformatics, who was not involved in the selection and allocation of patients, prepared coded, sealed envelopes that were opened in the presence of the patient to learn the treatment allocation. Treatment was planned within the subsequent four weeks.

Figure 1.

Needle placement and depiction of herniation size.

Intervention

Conventional surgery consisted of a microdiscectomy performed under general or spinal anaesthesia by surgeons with extensive experience in the technique. The aim of the surgery was to remove the herniated disc fragment, using a unilateral transflaval approach. Patients were operated with loupe magnification or microscope depending on the surgeon’s preference. The length of hospital stay was according to the usual care in the participating hospitals.

In the PLDD group, treatment was performed with the patient in prone position using a combination of computed tomography (CT) and fluoroscopy guidance. An 18G needle was placed centrally in the nucleus pulposus (Figure 1, Table 1) and parallel to the endplates by means of a posterolateral approach under local anaesthesia and sterile conditions (Figure 2). This technique is exactly the same as a routine discography. Through the needle, a glass fibre of 600 micron was advanced into the disc, enabling the application of laser energy (Diode laser, Biolitec, 980 nm, 7 W, 0.6 s pulses, interval 1 s) to a total energy delivered of 1500 J (2000 J for level L4–5). The treatment took place in an outpatient setting and was performed by an interventional (neuro-)radiologist, with ample experience in these spinal procedures.

Table 1.

Inclusion and exclusion criteria.

Inclusion:

Age 18–70 years

Persistent radicular pain lasting more than 6–8 weeks

Operation indication

Disc herniation confirmed at MRI

Unilateral disc herniation smaller or equal to 1/3 of the  spinal canal

Informed consent

Exclusion:

Previous surgery at the same disc level

Cauda equina syndrome

Spondylolytic or degenerative spondylolisthesis

Central spinal canal sentosis

Pregnancy

Severe somatic or psychiatric illness

Planned (e)migration to another country in the year after the  inclusion

Inadequate verbal or writing skills in Dutch language

MRI: magnetic resonance imaging.

Figure 2.

(a) Position of needle in the disc, with gas formation after percutaneous laser disc decompression (PLDD). (b) Procedural image with the C-arch and computed tomography (CT) set-up during the study.

Both treatment strategies were followed by active mobilisation in the post-operative period. Early resumption of daily activities and work were encouraged in both groups.

Outcome measures

The primary outcome measure used was the patient’s self-reported functional disability in the modified Roland-Morris Disability Questionnaire (RDQ) for sciatica, which was used for the power calculation. For primary endpoint analyses we also used the perceived recovery with regard to leg pain on a seven-point Likert scale and the 100 mm visual analogue scale (VAS) for leg and back pain.^13–15 These scores were collected at 1, 3, 4, 6, 8, 12, 26, 38, 52, 78 and 104 weeks after randomisation. Secondary outcome measures, at 4, 8, 26, 52 and 104 weeks, were the functional and economic scores on the Prolo scale, bodily pain and physical functioning scores on the Short Form-36 (SF-36), Sciatica Frequency and Bothersomeness index scores, neurological examination by the research nurse and complication- and reoperation rates.^13,16–18 PLDD treatment is nonsurgical and performed without skin incision, without general/spinal anaesthesia and in the radiology department as an outpatient procedure, therefore patients and research nurses could not be blinded for the randomised strategy.

Statistical analysis

The study was initially designed as an equivalence study comparing percentages of success on the RDQ scale. Because of an unexpected low accrual, the design was modified before the data were analysed in any respect. Our protocol was amended in two respects: Firstly the outcome measure was reverted to the underlying continuous RDQ scale to increase power and secondly the main statistical test was changed to a non-inferiority design (instead of a symmetric equivalence design). In our study, a difference of four points on the RDQ has been used as the minimum clinically important difference as long as the mean improvement is at least 11 points.¹⁹ The power calculation is performed under the alternative hypothesis of a difference of one point on the RDQ scale. By using the value four as the upper limit of the equivalence interval, with a-level of 0.05, a power of 0.90 and a standard deviation (SD) = 5, the required sample size was 98 (49 per treatment arm). To adjust for an estimated 8% loss to follow-up, we planned to include at least 110 patients. The calculations were based on a comparison of averages at a single point in time (week 52). By using a repeated measurements analysis of variance, the power of the study over the entire follow-up period was assumed to be substantially higher (in case of non-crossing curves) or equivalently, the average difference between the treatments over time (or area under the curve (AUC)) can be established even more accurately.

Baseline data were compared between the two treatment groups using the Chi-square or Student t-test, as appropriate.

When analysing the RDQ as primary outcome measure, the primary analysis consists of comparisons at week 4, 8, 26, 52 and 104 (point estimates) and a repeated measurement analysis of variance (comparing the averages over the entire two-year period).

In the analysis of the seven-point Likert scale of perceived recovery for leg-pain compared to baseline, the scores are dichotomised defining recovery as ‘complete’ or ‘nearly complete’.

The secondary outcome measures were also assessed in a repeated measurements analysis of variance. The scores were expressed as means with 95% confidence intervals (CIs). Point-wise estimates and their CI were obtained by using models with time as a categorical covariate to allow assessment of systematic patterns. The differences between randomisation groups were determined by estimating the main treatment effect and the treatment-time interaction over the two-year period. Non-inferiority is concluded at two-year follow-up resp. over the entire two-year interval if the 95% CI of the treatment effect at that point respectively over that period excludes the pre-established boundary of four points.

Patients were not excluded from the repeated-measures analysis of variance, even when one or more observations for a patient were missing at specific time points but assumed to be missing at random.

All data were stored via the Internet-based secure data management system (ProMISe) of the department of Medical Statistics and Bioinformatics. Analyses were carried out by PB and RB using SPSS software version 17 (SPSS Inc., Chicago, Illinois, USA).

Results

Patient characteristics

Between January 2005–September 2007 we included 115 patients for the trial. Fifty-five patients were allocated to PLDD and 57 to surgery. Three patients were excluded after randomisation (Figure 3). Baseline characteristics showed no significant differences between both treatment groups (Table 2).

Figure 3.

Flow chart of patient inclusion.

Table 2.

Patient characteristics at baseline.

		PLDD		Surgery
Parameter		Mean		Mean
Age	Mean (SE), year	43.2	11.8	43.7	9.7
Female sex	No (%)	19	35	24	42
BMI	Mean (SE)	25.1	4.4	25	5.2
Smoker	No (%)	23	42	18	32
Duration of sciatica	Median (range weeks)	30	9–182	26	8–260
8–26 weeks	No (%)	12	23	26	46
26–52 weeks	No (%)	21	40	12	21
52+ weeks	No (%)	19	37	18	32
Sick leave	No (%)	26	49	31	55
Radicular pain right leg	No (%)	28	51	27	47
Sensory disturbance	No (%)	46	100	45	94
Muscle weakness	No (%)	33	40	34	60
Asymmetric tendon reflexes
Knee	No (%)	5	9	7	13
Ankle	No (%)	6	11	15	27
Pain
SLR	No (%)	47	86	51	90
XSLR	No (%)	9	17	15	27
Slump	No (%)	42	79	43	80
HNP level
L2–3	No (%)	1	2	0	0
L3–4	No (%)	5	9	3	5
L4–5	No (%)	21	39	25	44
L5–S1	No (%)	26	48	28	49
Other	No (%)	1	2	1	2
Roland Disability Questionnaire	Mean (SE)	15.7	4.9	15.5	4.7
VAS score
Back pain	Mean (SE)	44.7	27.6	45.8	26.7
Leg pain		56.9	20.4	60.7	19.9
General health		47.3	24.7	49.4	23.7
Prolo scale
Function	Mean (SE)	1.1	0.6	0.9	0.5
Economic		1.7	1.7	2.1	1.7
Rand SF-36
Bodily pain	Mean (SE)	32.8	20.5	30	16.1
Physical functioning		41	22.6	38.6	20.9
Sciatica index
Frequency	Mean (SE)	3.6	1.1	3.8	1.2
Bothersomeness		3.3	1.2	3.1	1.3
Preference for treatment
PLDD	No (%)	17	32	19	34
No preference		23	43	30	54
Surgery		11	20	6	11
Time to surgery	Mean (SE), days	13.9	8.8	18.6	9

BMI: body mass index; PLDD: percutaneous laser disc decompression; SE: standard error; SF-36: Short Form 36; SLR: straight leg raising; VAS: visual analogue scale; XSLR: crossed straight leg raising.

At two-year follow-up, three patients (5%) in the surgery group and four patients (7%) in the PLDD group were lost to follow-up. Hence, data from 51 patients (88%) in the surgery group were analysed for the primary outcome measure, against 48 patients (87%) in the PLDD group (Figure 3).

At two years, four patients (8%) had received surgery after a technically failed PLDD procedure, and 25 patients (45%) underwent additional surgery after technically successful PLDD treatment but without subsequent clinical recovery. A total of 66 re-operations were performed in those 29 patients (range 1–3 reoperations). In the conventional surgery group, twelve patients (21%) needed repeated surgery with a total of 26 re-operations (range 1–2 reoperations). The relative risk of having a re-operation in the PLDD group is 2.5. In 13 patients, evenly spread over the two groups (n = 6 and n = 7), re-operations were performed after initial recovery following the initial treatment.

Treatment effects

Table 3 shows the scores of the primary outcome measures over the two-year follow-up period. The score on the 24-point Roland Morris Disability Questionnaire improved from a mean score of 15.7 and 15.5, in the PLDD and surgery group respectively, to 4.3 and 4.4 at two years after randomisation. Since the 95% CI is (–1.3 – +1.3) and therefore excludes the non-inferiority boundary, PLDD can be classified as ‘non-inferior’; actually since both endpoints are well below the value of four, equivalence of the two strategies can be inferred with respect to the Roland score. (Figure 4) The analysis of the RDQ showed no significant difference between the two groups at the two-year follow-up. The mean group difference, over the entire 104-week period, was also not significant. The VAS scores for leg and back pain also showed improvement in both groups without significant difference at the predefined time points (Figures 5 and 6). The between-group-difference at two years was also not statistically significant.

Table 3.

Postoperative outcomes at 0–104 weeks.

	PLDD		Surgery				Treatment effect	Treatment× time interaction
Outcome measure	Mean	SE	Mean	SE	Difference between treatments 95% CI		p-value	p-value
Roland Disability Questionnaire
1–104 weeks	8.2	0.5	8.2	0.5	0.0	–1.3–1.3	1.00	0.06
Week 4	10.8	0.8	13.2	0.7	2.5	0.2–4.7
Week 8	7.8	0.9	7.8	0.7	–0.1	–2.3–2.1
Week 26	7	1	4.8	0.8	–2.2	–4.4–0.1
Week 52	5.4	1	4.4	0.7	–1.1	–3.4–1.1
Week 104	4.3	0.9	4.4	0.8	0.1	–2.2–2.4
VAS-leg pain
1–104 weeks	24.8	1.9	19.8	1.9	–5.0	–10.2–0.2	0.06	0.42
Week 4	38.7	3.9	30.9	3.6	–7.4	–16.8–1.9
Week 8	24.9	3.5	20.1	3.2	–5.7	–15.0–3.7
Week 26	19.9	3.8	15.7	3.3	–4.2	–13.6–5.2
Week 52	18.1	3.1	12.6	2.4	–5.7	–15.2–3.8
Week 104	11.1	3.5	13.8	3.4	2.7	–6.8–12.3
VAS-back pain
1–104 weeks	28.8	1.8	25.9	1.9	–3.0	–8.1–2.2	0.26	0.58
Week 4	38	3.5	39.7	3.7	2	–7.2–11.3
Week 8	28.7	3.6	22.6	2.8	–6.3	–15.5–2.9
Week 26	30.6	3.9	20.4	3.3	–9.4	–18.6–0.1
Week 52	22.9	3.3	16.6	2.6	–7.6	–16.9–1.7
Week 104	19.3	3.5	20.8	3.4	1.5	–8.0–11.00
General health
1–104 weeks	67.3	1.7	72.0	1.6	4.7	0.2–9.3	0.04	0.04
Week 4	63.8	3.3	58.4	3.5	–5.9	–14.8–3.1
Week 8	66.7	3.5	74.6	2.9	8.3	–0.6–17.2
Week 26	69.4	3.7	78.7	3	8.9	–0.2–17.9
Week 52	77	3.1	78.5	3	1.7	–7.3–10.8
Week 104	73.1	3.4	75	3.3	1.9	–7.5–11.4
SF-36 Pain
1–104 weeks	61.6	1.9	61.9	1.9	0.3	–5.0–5.6	0.91	0.14
Week 4	39.5	3	35.9	2.4	–4.1	–12.9–4.8
Week 8	51.1	3.3	50.7	2.9	–0.6	–9.3–8.1
Week 26	59.5	4.1	71.6	3.3	11.3	2.4–20.1
Week 52	70	3.4	72.4	3	2.5	–6.4–11.3
Week 104	75.3	3.3	73.1	3.3	–2.2	–11.4–7.0
SF-36 Physical Functioning
1–104 weeks	73.7	2.1	67.5	2	–6.1	–11.7––0.5	0.03	0.001
Week 4	59.4	3	41.2	2.7	–18.4	–26.8––10.0
Week 8	67.5	3.4	62.1	2.6	–5.6	–13.9–2.7
Week 26	70.8	3.5	74.3	3.2	3.2	–5.1–11.6
Week 52	77.8	3.2	81.2	2.7	3.2	–5.2–11.6
Week 104	81.6	3.2	77.3	3.1	–4.3	–13.2–4.5
Sciatica frequency
1–104 weeks	1.7	0.1	1.7	0.1	0.2	–0.3–0.7	0.92	0.92
Week 4	2.3	0.2	2.3	0.2	–0.1	–0.6–0.4
Week 8	1.8	0.2	1.6	0.1	–0.1	–0.5–0.4
Week 26	1.6	0.2	1.5	0.2	–0.1	–0.6–0.4
Week 52	1.5	0.2	1.3	0.2	–0.1	–0.6–0.3
Week 104	1.3	0.2	1.5	0.2	0	–0.3–0.4
Sciatica bothersome
1–104 weeks	1.4	0.1	1.3	0.1	–0.1	–0.4–0.2	0.56	0.89
Week 4	2.0	0.2	1.9	0.2	–0.1	–0.6–0.4
Week 8	1.6	0.2	1.3	0.1	–0.2	–0.7–0.2
Week 26	1.3	0.2	1.0	0.2	–0.3	–0.7–0.2
Week 52	1.2	0.2	1.0	0.1	–0.2	–0.7–0.2
Week 104	0.9	0.2	1.0	0.2	0.1	–0.4–0.6

CI: confidence interval; PLDD: percutaneous laser disc decompression; SE: standard error; SF-36: Short Form 36.

Figure 4.

Roland Disability Scale Score at 0–104 weeks.

Figure 5.

Visual analogue score (VAS) for leg pain at 0–104 weeks.

Figure 6.

Visual analogue score (VAS) for back pain at 0–104 weeks.

The endpoint analysis of the perceived recovery scale showed no significant difference at two years; 70.8% of patients in the PLDD group showed recovery versus 60.8% in the surgery group (odds ratio 1.6 (95% CI 0.7–3.6)) (Figures 7 and 8).

Figure 7.

Perceived recovery according to the Likert scale at 0–104 weeks.

Figure 8.

Percentage of patients that perceived recovered at 0–104 weeks.

However, the time to first recovery during the follow-up period was attained at a significantly slower rate in the PLDD group, resulting in a hazard ratio of 0.64 (95% CI 0.43–0.96) compared to primary surgery.

The General Health item on the SF-36 scale scored significantly better over the 1–104 week follow-up period in the surgery group compared to PLDD. The mean difference was 4.7 points (95% CI 0.2–9.3), which is without clinical implications. The Physical Functioning score on the SF-36 scale was significantly better in the PLDD group over the 1–104 week follow-up period with a mean difference of –6.1 points (95% CI –11.7 to –0.5). The largest difference was found at four-weeks follow-up. At 26 and 52 weeks the score was slightly better in the surgery group, changing to a better result again in the PLDD group at 104 weeks. All other outcome measures showed non-significant differences.

Discussion

Although the sample size of this study is small, the results show that a treatment strategy of PLDD, followed by surgery when needed, resulted in similar outcomes compared to surgery alone at both the one and two-year follow-up period. In this study we did observe some unexpected phenomena regarding perceived recovery and reoperation rate:

First, the observed recovery rate in the surgery group (60.8%) is low compared to other trials reporting around 80% recovery.^3,5

Second, in the time between randomisation and follow-up, a large proportion of the patients in both groups underwent reoperation. Thirteen patients, with an equal distribution between the groups, had a reoperation after initial recovery, which can therefore be attributed to recurrence of the symptoms. The reoperation rate in the PLDD group was 45% due to ineffective PLDD treatment and an additional 7% was caused by a failure to perform PLDD and hence conversion to surgery. In the surgery group the reoperation rate was 21%, which is high compared to the 2–10% reoperation rates for surgery quoted in literature.^5,20

Third, the percentage of patients considering themselves recovered, based on the Likert scale, was 60.8% in the surgery group versus 70.8% in the PLDD group at two-year follow-up. In the surgery group this percentage is lower than the one-year result of 75%, whereas recovery in the PLDD group increased from 69%. A previous publication of Barth et al. also pointed to this phenomenon of worsening perceived recovery at two years after microdiscectomy.^21,22

One explanation for these three observations may be that the patients in our trial had small and contained disc herniations, which are known to have a relatively poor response to both surgical and conservative treatment.^23,24 The choice for this subgroup of patients in this study was intentional since, at the time of design of the trial, the contained herniations were considered most appropriate for PLDD treatment and other minimally invasive therapies.^12,25,26 The fact that, in a parallel trial, comparing microdiscectomy with microtube surgery performed by the same neurosurgeons, we showed a lower reoperation rate within the normally reported range, indicates that the observed results were not caused by inexperienced surgeons but more probably by the fact that the smaller herniated discs do have more dubious indications for surgery as nerve root compression is less clear and other local factors may be of more importance.²⁷

Another explanation may be that in our population a large cohort of patients had a long history of sciatica, with 37% of the PLDD group and 32% of the surgery group experiencing complaints for more than one year. This subgroup of patients is also known to have poorer outcomes after surgery than patients with a short duration of complaints.³⁰ We assume that this will impact both treatment groups in a similar way.

For the PLDD population, it could be that the herniated disc has desiccated over time. A disc that contains less fluid will likely respond less to laser energy being applied to vaporise the nucleus pulposus. If this is the case, PLDD treatment is likely to be more effective in vaporising fresh herniated discs, which would have to be studied in a different study.

One added explanation for the higher reoperation rate in the PLDD group is the ‘technical failure' in four patients. In three of those cases we failed to perform the treatment due to factors such as osteophytes or syndesmophytes. Since the inclusion was done by the neurosurgeon, to avoid ‘cherry picking' by the treating interventionalist, those local factors that prohibit puncture may not have been recognised prior to treatment. Furthermore, in one case the laser apparatus did not function and the procedure had to be aborted after the needle was inserted. Since repeated treatment was not allowed in the protocol, the patient was offered open surgery.

One more potential cause of having worse results in both groups can be that the including neurosurgeons were forced to choose between this trial and another running simultaneously. The latter study on microendoscopic techniques included patients with larger disc herniations, with more apparent nerve root compression. For patients with small disc herniations, the PLDD trial was an alternative. If a neurosurgeon had any doubts as to go ahead with surgery or not in those more difficult cases, it would be conceivable that those patients were offered the trial to avoid open surgery in 50%.

The General Health item on the SF-36 scale scored significantly better over the entire two-year follow-up period in the surgery group compared to PLDD. The mean difference of 4.7 (95% CI 0.2–9.3) points around the score of 70 on a 100 point scale, is clinically not considered relevant.³¹

The Physical Functioning score on the SF-36 scale scored significantly better over the follow-up period too. A large part of this observation may be explained by the large difference at the four-week time point where the difference, in favour of the PLDD group, is clinically relevant.³¹ However, at all the other follow-up periods the difference between the groups is without clinical importance.

Conclusion

This first randomised controlled trial comparing PLDD and microdiscectomy demonstrated that PLDD has a similar outcome to microdiscectomy over a two-year follow-up period.

Both groups in our study had an unexpectedly high reoperation rate, which may be attributable to the selection of patients with small herniations and long standing complaints. After this study there seems to be a justification to perform a study into the value of PLDD as an alternative to conservative treatment in the early phase.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Dutch Health Care Insurance Board. The Dutch Health Care Insurance Board did not participate in the design and conduct of the study, in the collection, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.